Segmentation 101 : The Largest list of Open and Searchable Segmentation Datasets

Segmentation 101 a.k.a. SEG 101 is an exhaustive list we have created to make it easier for you to search publicly available Image Segmentation datasets.

Segmentation is one of the most time-consuming annotation tasks. Sometimes, before collecting your own dataset, you do want to experiment on a publicly available dataset.

So, we created this list which is searchable by class name, so you can quickly find a class that you need. It contains Instance Segmentation, Semantic Part Segmentation, Motion Segmentation, Vessel Segmentation, and many such variants.

So, if you want a head start for your AI app or for a hobby project which requires pixel-wise annotated data then, do make a quick stop at the below-mentioned datasets, and use them to build something great !!.


1. The COCO-Stuff Dataset

    Dataset Characteristics:

Task Type: Instance Segmentation, Scene Understanding, Object Localization, Semantic Segmentation.

Image Count: 164K complex images from COCO 2017

Labeled Instances Count: 2.5 million

Categories: 172 classes: 80 things, 91 stuff, and 1 class unlabeled

    *91 thing classes (1-91) Descriptions for stuff labels (92-182) are provided. Furthermore, 11 classes have been removed from COCO and therefore lack a preview image.

Id Label name Preview Description
0 unlabeled Pixels that do not belong to any of the other classes
1 person (view)
2 bicycle (view)
3 car (view)
4 motorcycle (view)
5 airplane (view)
6 bus (view)
7 train (view)
8 truck (view)
9 boat (view)
10 traffic light (view)
11 fire hydrant (view)
12 street sign Removed from COCO.
13 stop sign (view)
14 parking meter (view)
15 bench (view)
16 bird (view)
17 cat (view)
18 dog (view)
19 horse (view)
20 sheep (view)
21 cow (view)
22 elephant (view)
23 bear (view)
24 zebra (view)
25 giraffe (view)
26 hat Removed from COCO.
27 backpack (view)
28 umbrella (view)
29 shoe Removed from COCO.
30 eye glasses Removed from COCO.
31 handbag (view)
32 tie (view)
33 suitcase (view)
34 frisbee (view)
35 skis (view)
36 snowboard (view)
37 sports ball (view)
38 kite (view)
39 baseball bat (view)
40 baseball glove (view)
41 skateboard (view)
42 surfboard (view)
43 tennis racket (view)
44 bottle (view)
45 plate Removed from COCO.
46 wine glass (view)
47 cup (view)
48 fork (view)
49 knife (view)
50 spoon (view)
51 bowl (view)
52 banana (view)
53 apple (view)
54 sandwich (view)
55 orange (view)
56 broccoli (view)
57 carrot (view)
58 hot dog (view)
59 pizza (view)
60 donut (view)
61 cake (view)
62 chair (view)
63 couch (view)
64 potted plant (view)
65 bed (view)
66 mirror Removed from COCO.
67 dining table (view)
68 window Removed from COCO.
69 desk Removed from COCO.
70 toilet (view)
71 door Removed from COCO.
72 tv (view)
73 laptop (view)
74 mouse (view)
75 remote (view)
76 keyboard (view)
77 cell phone (view)
78 microwave (view)
79 oven (view)
80 toaster (view)
81 sink (view)
82 refrigerator (view)
83 blender Removed from COCO.
84 book (view)
85 clock (view)
86 vase (view)
87 scissors (view)
88 teddy bear (view)
89 hair drier (view)
90 toothbrush (view)
91 hair brush Removed from COCO.
92 banner (view) Any large sign, especially if constructed of soft material or fabric, often seen in stadiums and advertising.
93 blanket (view) A loosely woven fabric, used for warmth while sleeping.
94 branch (view) The woody part of a tree or bush, arising from the trunk and usually dividing.
95 bridge (view) A manmade construction that spans a divide (incl. train bridge, river bridge).
96 building-other (view) Any other type of building or structures.
97 bush (view) A woody plant distinguished from a tree by its multiple stems and lower height (incl. hedge, scrub).
98 cabinet (view) A storage closet, often hanging on the wall.
99 cage (view) An enclosure made of bars, often seen in zoos.
100 cardboard (view) A wood-based material resembling heavy paper, used in the manufacture of boxes, cartons and signs.
101 carpet (view) A fabric used as a floor covering.
102 ceiling-other (view) Other types of ceilings (incl. industrial ceilings, painted ceilings).
103 ceiling-tile (view) A ceiling made of regularly-shaped slabs.
104 cloth (view) A piece of cloth used for a particular purpose. (incl. cleaning cloth).
105 clothes (view) Items of clothing or apparel, not currently worn by a person.
106 clouds (view) A visible mass of water droplets suspended in the air.
107 counter (view) A surface in the kitchen or bathroom, often built into a wall or above a cabinet, which holds the washbasin or surface to prepare food.
108 cupboard (view) A piece of furniture used for storing dishware or a wardrobe for clothes, sometimes hanging on the wall.
109 curtain (view) A piece of cloth covering a window, bed or shower to offer privacy and keep out light.
110 desk-stuff (view) A piece of furniture with a flat surface and typically with drawers, at which one can read, write, or do other work.
111 dirt (view) Soil or earth (incl. dirt path).
112 door-stuff (view) A portal of entry into a building, room or vehicle, consisting of a rigid plane movable on a hinge (incl. the frame, replaces door).
113 fence (view) A thin, human-constructed barrier which separates two pieces of land.
114 floor-marble (view) The supporting surface of a room or outside, made of marble.
115 floor-other (view) Any other type of floor (incl. rubber-based floor).
116 floor-stone (view) The supporting surface of a room or outside, made of stone (incl. brick floor).
117 floor-tile (view) The supporting surface of a room or outside, made of regularly-shaped slabs (incl. tiled stone floor, tiled marble floor).
118 floor-wood (view) The supporting surface of a room or outside, made of wood (incl. wooden tiles, parquet, laminate, wooden boards).
119 flower (view) The seed-bearing part of a plant (incl. the entire flower).
120 fog (view) A thick cloud of tiny water droplets suspended in the atmosphere near the earth’s surface.
121 food-other (view) Any other type of food.
122 fruit (view) The sweet and fleshy product of a tree or other plant.
123 furniture-other (view) Any other type of furniture (incl. oven).
124 grass (view) Vegetation consisting of typically short plants with long, narrow leaves (incl. lawn, pasture).
125 gravel (view) A loose aggregation of small water-worn or pounded stones.
126 ground-other (view) Any other type of ground found outside a building.
127 hill (view) A naturally raised area of land, not as high as a mountain, viewed at a distance and may be covered in trees, snow or grass.
128 house (view) A smaller size building for human habitation.
129 leaves (view) A structure of a higher plant, typically green and blade-like, that is attached to a stem or stalk.
130 light (view) A source of illumination, especially a lamp (incl. ceiling lights).
131 mat (view) A piece of coarse material placed on a floor for people to wipe their feet on.
132 metal (view) A raw metal material (incl. a pile of metal).
133 mirror-stuff (view) A glass coated surface which reflects a clear image (incl. the frame, replaces mirror).
134 moss (view) A small flowerless green plant which lacks true roots, growing in in damp habitats.
135 mountain (view) A large natural elevation rising abruptly from the surrounding level, viewed at a distance and may be covered in trees, snow or grass.
136 mud (view) A soft, sticky matter resulting from the mixing of earth and water.
137 napkin (view) A piece of cloth or paper used at a meal to wipe the fingers or lips.
138 net (view) An open-meshed fabric twisted, knotted, or woven together at regular intervals.
139 paper (view) A material manufactured in thin sheets from the pulp of wood.
140 pavement (view) A typically raised paved path for pedestrians at the side of a road.
141 pillow (view) A rectangular cloth bag stuffed with soft materials to support the head.
142 plant-other (view) Any other type of plant.
143 plastic (view) Raw plastic material.
144 platform (view) A raised level surface on which people or things can stand (incl. railroad platform).
145 playingfield (view) A ground marked off for various games (incl. indoor and outdoor).
146 railing (view) A fence or barrier made of typically metal rails.
147 railroad (view) A track made of steel rails along which trains run (incl. the wooden beams).
148 river (view) A stream of flowing water.
149 road (view) A paved way leading from one place to another.
150 rock (view) The solid mineral material forming part of the surface of the earth.
151 roof (view) The structure forming the upper covering of a building.
152 rug (view) A floor covering of thick woven material, typically not extending over the entire floor.
153 salad (view) A cold dish of various mixtures of raw or cooked vegetables.
154 sand (view) A loose granular substance, typically pale yellowish brown, resulting from erosion (incl. beach).
155 sea (view) Expanse of water that covers most of the earth’s surface.
156 shelf (view) An open piece of furniture that provides a surface for the storage or display of objects.
157 sky-other (view) Any other type of sky (incl. blue sky).
158 skyscraper (view) A very tall building of many storeys.
159 snow (view) Atmospheric water vapour frozen into ice crystals, falling or lying on the ground.
160 solid-other (view) Any other type of solid material.
161 stairs (view) A set of steps leading from one floor to another (incl. stairs inside or outside a building).
162 stone (view) A piece of stone shaped for a purpose.
163 straw (view) Dried stalks of grain.
164 structural-other (view) Any other type of structural connection (incl. arcs, pillars).
165 table (view) A piece of furniture with a flat top and one or more legs.
166 tent (view) A portable shelter made of cloth.
167 textile-other (view) Any other type of textile.
168 towel (view) A piece of thick absorbent cloth used for drying oneself.
169 tree (view) A woody plant, typically having a single trunk growing to a considerable height and bearing lateral branches at some distance from the ground.
170 vegetable (view) A part of a plant used as food.
171 wall-brick (view) A building wall made of bricks of clay.
172 wall-concrete (view) A building wall made of concrete.
173 wall-other (view) Any other type of wall.
174 wall-panel (view) A panel that is attached to a wall.
175 wall-stone (view) A building wall made of stone.
176 wall-tile (view) A building wall made of tiles, such as used in bathrooms and kitchens.
177 wall-wood (view) A building wall made of wooden material.
178 water-other (view) Any other type of water (incl. lake).
179 waterdrops (view) Sprinkles or drops of water not connected to a larger body of water.
180 window-blind (view) Blinds and shutters that cover a window.
181 window-other (view) Any type of window that must be visible in the image (replaces window).
182 wood (view) Raw wood materials (incl. logs).

Code/Model/Publication URLCOCO-Stuff: Thing and Stuff Classes in Context

Source URL : The COCO-Stuff dataset

Dataset Examples :

2. The DUTS Image Dataset

     Dataset Characteristics:

     Task Type : Saliency detection

     Image Attribute : 10,553 training images and 5,019 test images

    Code/Model/Publication URL : Learning to Detect Salient Objects with Image-level Supervision

    Source URL : The DUTS Image Dataset

    Dataset Examples :

 

3. The CMU-Cornell iCoseg Dataset

     Dataset Characteristics:

     Task Type : cosegmentation

     Image Attribute : 38 challenging groups with 643 total images (∼17 images per group), with   associated pixel-level ground truth.

    Code/Model/Publication URL : iCoseg: Interactive Co-segmentation with Intelligent Scribble Guidance

    Source URL :   The CMU-Cornell iCoseg Dataset

    Dataset Examples :

 

4. The PASCAL-S Dataset

    Dataset Characteristics:

     Task Type : fixation prediction, salient object segmentation

     Image Attribute : 850 images from PASCAL VOC 2010

     Object Instance : 1296

     No of subjects : 12

    Code/Model/Publication URL : The Secrets of Salient Object Segmentation

    Source URL :   The PASCAL-S Dataset

    Dataset Examples :

Credits : The PASCAL-S Dataset

 

5. DAVIS: Densely Annotated Video Segmentation

     Dataset Characteristics:

Task Type: video object segmentation

Video Attribute: 400 objects on 150 videos with∼10k frames

Code/Model/Publication URL: A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation

Source URL: DAVIS

Dataset Examples :

 

6. NYU Depth Dataset V1

    Dataset Characteristics:

Task Type : Indoor Scene segmentation, Multi-class segmentation

Image Attribute : 2347 unique frames cover over 13 object classes, spread over 64 different indoor environments.

Scene class

Scenes

Bathroom 6
Bedroom 17
Bookstore 3
Cafe 1
Kitchen 10
Living Room 13
Office 14
Total 64
Object Class
Bed
Blind
Bookshelf
Cabinet
Ceiling
Floor
Picture
Sofa
Table
Television
Wall
Window
Background

Code/Model/Publication URL : Indoor Scene Segmentation using a Structured Light Sensor

Source URL : NYU Depth V1

Dataset Examples :

Credits : Indoor Scene Segmentation using a Structured Light Sensor

 

7. PASCAL-Part Dataset

    Dataset Characteristics:

Task Type: Body parts segmentation

Image Attribute: 10, 103

Label Attribute: 4,203

Categories: 20 with their individual body parts

Objects Parts
aeroplane
body
engine*
left wing
right wing
stern
tail
wheel*
bicycle
back wheel
chain wheel
front wheel
handlebar
headlight*
saddle
bird
beak
head
left eye
left foot
left leg
left wing
neck
right eye
right foot
right leg
right wing
tail
torso
boat
bottle
body
cap
bus
back license plate
back side
door*
front license plate
front side
headlight*
left mirror
left side
right mirror
right side
roof side
wheel*
window*
car
back license plate
back side
door*
front license plate
front side
headlight*
left mirror
left side
right mirror
right side
roof side
wheel*
window*
cat
head
left back leg
left back paw
left ear
left eye
left front leg
left front paw
neck
nose
right back leg
right back paw
right ear
right eye
right front leg
right front paw
tail
torso
chair
cow
head
left back lower leg
left back upper leg
left ear
left eye
left front lower leg
left front upper leg
left horn
muzzle
neck
right back lower leg
right back upper leg
right ear
right eye
right front lower leg
right front upper leg
right horn
tail
torso
diningtable
dog
head
left back leg
left back paw
left ear
left eye
left front leg
left front paw
muzzle
neck
nose
right back leg
right back paw
right ear
right eye
right front leg
right front paw
tail
torso
horse
head
left back hoof
left back lower leg
left back upper leg
left ear
left eye
left front hoof
left front lower leg
left front upper leg
muzzle
neck
right back hoof
right back lower leg
right back upper leg
right ear
right eye
right front hoof
right front lower leg
right front upper leg
tail
torso
motorbike
back wheel
front wheel
handlebar
headlight*
saddle
person
hair
head
left ear
left eye
left eyebrow
left foot
left hand
left lower arm
left lower leg
left upper arm
left upper leg
mouth
neck
nose
right ear
right eye
right eyebrow
right foot
right hand
right lower arm
right lower leg
right upper arm
right upper leg
torso
pottedplant
plant
pot
sheep
head
left back lower leg
left back upper leg
left ear
left eye
left front lower leg
left front upper leg
left horn
muzzle
neck
right back lower leg
right back upper leg
right ear
right eye
right front lower leg
right front upper leg
right horn
tail
torso
sofa
train
coach back side*
coach front side*
coach left side*
coach right side*
coach roof side*
coach*
head
head back side
head front side
head left tside
head right side
head roof side
headlight*
tvmonitor
screen

(* denotes a part can appear multiple times in one object instance)

Code/Model/Publication URL : Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Parts

Source URL : PASCAL-Part Dataset

Dataset Examples :

Credits: PASCAL-Part Dataset

 

8. Matting Human Datasets

    Dataset Characteristics:

Task Type: Human segmentation

Image Attribute: 34,427 images and corresponding matting results

Code/Model/Publication URL: matting human dataset

Source URL: matting_human_half

          Dataset Examples :

 

9. Automatic Portrait Segmentation Dataset

   Dataset Characteristics:

Task Type: Automatic Portrait Segmentation for styling images

Image Attribute: 1800

Code/Model/Publication URL : Automatic Portrait Segmentation for Image Stylization

Source URL: Automatic Portrait Segmentation 

Dataset Examples :

Credits : Automatic Portrait Segmentation 

 

10. ADE20K Dataset

   Dataset Characteristics:

Task Type: image segmentation

Image Attribute: 22,210 images,  900 scene categories defined in the SUN database, more than 200 object classes.

Label attribute: 22,210 Fully annotated with objects and parts

Code/Model/Publication URL: Semantic Understanding of Scenes through ADE20K Dataset

Source URL: ADE20K Dataset

Dataset Examples :

 

11. Pedestrian Parsing Dataset

      Dataset Characteristics:

Task Type: pedestrian detection, pose estimation, and body segmentation.

Image Attribute: contains 3,673 images from 171 videos of different Surveillance Scenes (PPSS), where 2,064 images are occluded and 1,609 are not.

Code/Model/Publication URL: Pedestrian Parsing via Deep Decompositional Network

Source URL: Pedestrian Parsing Dataset

Dataset Examples :

 

12. iSAID: A Large-scale Dataset for Instance Segmentation in Aerial Images

     Dataset Characteristics:

Task Type:  instance-level object detection and pixel-level segmentation task on aerial images

Image Attribute: 2806 high-resolution aerial images

Label attribute: 655,451 object instances

Categories :

iSAID Dataset Large-vehicle, swimming pool, Helicopter, Bridge, Plane, Ship, Soccer-ball field, Storage tank
Basketball court, Ground track field, Small-vehicle, Harbor, Baseball Diamond, Tennis court, Roundabout

Code/Model/Publication URL : iSAID: A Large-scale Dataset for Instance Segmentation in Aerial Images

Source URL: iSAID

Dataset Examples :

Credits: iSAID

 

13. Pascal VOC (2007-2012)

      Dataset Characteristics:

Task Type: object segmentation, class segmentation

Image Count: 54k Images

Labeled Instances Count: 19,377

Categories :

      • Person: person
      • Animal: bird, cat, cow, dog, horse, sheep
      • Vehicle: aeroplane, bicycle, boat, bus, car, motorbike, train
      • Indoor: bottle, chair, dining table, potted plant, sofa, TV/monitor
      • Background, void

Code/Model/Publication URLThe 2005 PASCAL Visual Object Classes Challenge

Source URL: Pascal VOC Dataset

Dataset Examples :

 

14. Open Images Dataset V6 

      Dataset Characteristics:

Task Type: object detection, segmentation, visual relationship, local narratives

Image Count: ~9M images

Labeled Instances Countsegmentation masks for 2.8M object instances in 350 classes, 16M bounding boxes in 600 object classes

Categories : 19,957 classes

Code/Model/Publication URL : Open Images

Source URL: Open Images Dataset 

Dataset Examples :

 

15. OASIS: A Large-Scale Dataset for Single Image 3D in the Wild

      Dataset Characteristics:

Task Type: Depth Estimation, Surface Normal Estimation, Fold & Occlusion detection, Planar Instance segmentation

Image Count : 140,000 images

Code/Model/Publication URLOASIS: A Large-Scale Dataset for Single Image 3D in the Wild

Source URL : OASIS: A Large-Scale Dataset for Single Image 3D in the Wild

Dataset Examples :

Credits : OASIS: A Large-Scale Dataset for Single Image 3D in the Wild

16. KAIST Salient Pedestrian Dataset

      Dataset Characteristics :

Task Type: pedestrian detector in thermal images

Image Attribute: 1702 images (913 day images and 789 night images)

Label instance: 4170 instances of pedestrians

Code/Model/Publication URL: Pedestrian Detection from Thermal Images using Saliency Maps

Source URL: Salient-Pedestrian-Detection

Dataset Examples :

Credits: Pedestrian Detection from Thermal Images using Saliency Maps

17. Clothing Attributes Dataset

      Dataset Characteristics :

Task Type: Dressing Style Analysis using Semantic Attributes

Image Attribute1856 images

Label instance : 283,107 label

          Categories : 26 attributes in total, including 23 binary-class attributes (6 for pattern, 11 for color and 6 miscellaneous attributes), 3 multi-class attributes (sleeve length, neckline shape and clothing category)

           Describing Clothing by Semantic Attributes

Clothing pattern (Positive / Negative) Solid (1052 / 441), Floral (69 / 1649), Spotted (101 / 1619)Plaid (105 / 1635), Striped (140 / 1534), Graphics (110 / 1668)
Major color(Positive / Negative) Red (93 / 1651), Yellow (67 / 1677), Green (83 / 1661), Cyan (90 / 1654)Blue (150 / 1594), Purple (77 / 1667), Brown (168 / 1576), White (466 / 1278)Gray (345 / 1399), Black (620 / 1124),>2Colors (203 / 1541)
Wearing necktie Yes 211, No 1528
Collar presence Yes 895, No 567
Gender Male 762, Female 1032
Wearing scarf Yes 234, No 1432
Skin exposure High 193, Low 1497
Placket presence Yes 1159, No 624
Sleeve length No sleeve (188), Short sleeve (323), Long sleeve (1270)
Neckline shape V-shape (626), Round (465), Others (223)
Clothing category Shirt (134), Sweater (88), T-shirt (108), Outerwear (220)Suit (232), Tank Top (62), Dress (260)

Code/Model/Publication URL : Describing Clothing by Semantic Attributes

Source URL : Clothing Attributes Dataset

Dataset Examples :

 

18. TACO: Trash Annotations in Context for Litter Detection

      Dataset Characteristics :

Task Type : litter detection and segmentation

Image Attribute : 2285 images

Label instance : 7426 annotations

           Categories: 60 categories which belong to 28 super (top) categories

Super category Category Notes
Aluminum foil Aluminum foil
Battery Battery
Blister pack Aluminum blister pack Containers used to store capsules (e.g. pills)
Blister pack Carded blister pack Paper-back package
Bottle Clear plastic bottle Water and soft drink bottles made of PET
Bottle Glass bottle Includes beer and wine bottles
Bottle Other plastic bottle Opaque or translucent. Generally made of HDPE. Includes detergent bottles
Bottle cap Plastic bottle cap
Bottle cap Metal bottle cap
Broken glass Broken glass
Can Aerosol
Can Drink can Aluminum soda can
Can Food can Steel can
Carton Corrugated carton Includes cardboard boxes
Carton Drink carton Tetrapak composites
Carton Egg carton
Carton Meal carton Includes sandwich boxes, paper plates, take-out boxes
Carton Pizza box
Carton Toilet tube
Carton Other carton Paperboard boxes
Cigarette Cigarette Cigarette butts
Cup Paper cup
Cup Disposable plastic cup Generally made of PET
Cup Foam cup Polystyrene Cup
Cup Glass cup
Cup Other plastic cup Reusable plastic cups, thicker than disposable ones
Food waste Food waste
Glass jar Glass jar
Lid Plastic lid Includes cup lids
Lid Metal lid Generally glass jar lids
Paper Normal paper
Paper Tissues
Paper Wrapping paper
Paper Magazine paper Plastified paper used in catalogues
Paper bag Paper bag Brown bag
Paper bag Plastified paper bag Bakery bags that come with transparent film
Plastic bag & wrapper Garbage bag
Plastic bag & wrapper Single-use carrier bag
Plastic bag & wrapper Polypropylene bag Reusable bags
Plastic bag & wrapper Plastic Film May be transparent or opaque. Inludes bread bags, cereal bags and produce bags
Plastic bag & wrapper Six pack rings
Plastic bag & wrapper Crisp packet So common that it needs its own category
Plastic bag & wrapper Other plastic wrapper Can be made of aluminium. Includes candy wrappers, retort pouches and yoghurt lids
Plastic container Spread tub Includes margarine tubs and yoghurt pots
Plastic container Tupperware HDPE microwavable tub
Plastic container Disposable food container Includes black trays and PET containers
Plastic container Foam food container Styrofoam takeaway boxes
Plastic container Other plastic container
Plastic glooves Plastic glooves
Plastic utensils Plastic utensils
Pop tab Pop tab
Rope Rope Includes fishing nets
Scrap metal Scrap metal Includes all metal except cans
Shoe Shoe
Squeezable tube Squeezable tube Includes toothpaste and glue tubes
Straw Plastic straw
Straw Paper straw
Styrofoam piece Styrofoam piece
Other plastic Other plastic Includes other objects or fragments made of plastic
Unlabeled litter Unlabeled litter Unknown object of unknown material. Any ambiguous object.

Code/Model/Publication URL : TACO: Trash Annotations in Context for Litter Detection

Source URL: TACO

Dataset Examples :

 

19. Crowd Instance-level Human Parsing (CIHP) Dataset

      Dataset Characteristics :

Task Type: Instance-level Human Parsing, Semantic Part Segmentation

Image Count: 38280 multiple-person images with pixel-wise annotations of 19 semantic parts and of 20 categories in instance-level.

Categories: pixel-wise annotations on 20 categories.

Semantic Part Label

Hat Pants
Hair Gloves
Sunglasses Scarf
Upper-clothes Skirt
Dress Torso-skin
Coat Face
Socks Right arm
Right leg Left shoe
Right shoe Left leg
Left arm

Code/Model/Publication URL: Instance-level Human Parsing via Part Grouping Network

Source URL: Crowd Instance-level Human Parsing (CIHP) Dataset

Dataset Examples :

Credits: Crowd Instance-level Human Parsing (CIHP) Dataset

20. ModaNet

      Dataset Characteristics :

Task Type: semantic segmentation on street fashion images in detail.

Image Count: 55,176 street image fully annotated with polygons.

CategoriesEach polygon (segmentation mask) annotation is assigned to one of the following labels:

Label Description Fine-Grained-categories
1 bag bag
2 belt belt
3 boots boots
4 footwear footwear
5 outer coat/jacket/suit/blazers/cardigan/sweater/Jumpsuits/Rompers/vest
6 dress dress/t-shirt dress
7 sunglasses sunglasses
8 pants pants/jeans/leggings
9 top top/blouse/t-shirt/shirt
10 shorts shorts
11 skirt skirt
12 headwear headwear
13 scarf & tie scarf & tie

Code/Model/Publication URL : ModaNet: A Large-scale Street Fashion Dataset with Polygon Annotations

Source URL: modanet

Dataset Examples :

Credits : ModaNet: A Large-scale Street Fashion Dataset with Polygon Annotations

 

21. Embrapa Wine Grape Instance Segmentation Dataset

Dataset Characteristics :

Task Type: instance segmentation for image-based monitoring and field robotics in viticulture.

Image Count: 300 images

            Label Count: 2,020 binary masks for instance segmentation

CategoriesEach polygon (segmentation mask) annotation is assigned to one of the following labels:

Prefix Variety
CDY Chardonnay
CFR Cabernet Franc
CSV Cabernet Sauvignon
SVB Sauvignon Blanc
SYH Syrah

Code/Model/Publication URL: Grape detection, segmentation and tracking using deep neural networks and three-dimensional association

Source URL : Embrapa WGISD

Dataset Examples :

Credits: Grape detection, segmentation and tracking using deep neural networks and three-dimensional association

 

22. MinneApple: A Benchmark Dataset for Apple Detection and Segmentation

Dataset Characteristics :

Task Type: fruit segmentation

Image Count: 1671 images of Apples

Label Count: 41,000 annotated object instances

Code/Model/Publication URL : MinneApple: A Benchmark Dataset for Apple Detection and Segmentation

Source URL: MinneApple

Dataset Examples :

Credits : MinneApple: A Benchmark Dataset for Apple Detection and Segmentation

23. FSS-1000: A 1000 Class Dataset for Few-shot Segmentation

      Dataset Characteristics :

Task Type: image segmentation

Image Count: 10000 images with pixelwise segmentation labels

Categories: 1,000,  with instance segmentation labels in 758 out of the 1,000 classes

Hierarchy of FSS-1000

Code/Model/Publication URL : FSS-1000: A 1000-Class Dataset for Few-Shot Segmentation

Source URL: FSS-1000

Dataset Examples :

Credits : FSS-1000: A 1000-Class Dataset for Few-Shot Segmentation

 

24. KINS Dataset

Dataset Characteristics :

Task Type: Amodal instance segmentation

Image Count: 14,991images from KITTI  dataset

Label Count: On average, each image has 12.53 labeled instances, and each object polygon consists of 33.70 points.

Categories :

Category People Vehicle
Subcategory Pedestrian

Cyclist

person-siting

Car

Van

Tram

truck

misc

Code/Model/Publication URL: Amodal Instance Segmentation with KINS Dataset

Source URL: KINS Dataset

Dataset Examples :

Credits: Amodal Instance Segmentation with KINS Dataset

25. MSeg

      Dataset Characteristics :

Task Type: semantic segmentation

Image Count : 80,000 images

Label Count: 220,000 object masks

Category: 194 categories.

Universal Class Name Description
airplane an aircraft that has a fixed wing and is powered by propellers or jets
animal-other Any animal that is not a bird, cat, dog, horse, sheep, cow, elephant, bear, zebra, or giraffe
apparel clothing in general, not currently being worn by a human being
apple fruit with red or yellow or green skin and sweet to tart crisp whitish flesh
arcade machine
armchair chair with a support on each side for arms (that is not a swivel chair, seat, or sofa)
autorickshaw in South Asia, a small motor vehicle with three wheels that is used as a taxi
awning a sheet of canvas or other material stretched on a frame and used to keep the sun or rain off a storefront, window, doorway, or deck. Should be attached to a building.
backpack a bag carried by a strap on your back or shoulder
bag
banana Elongated crescent-shaped yellow fruit with soft sweet flesh
banner long strip of cloth or paper used for decoration or advertising
barrel a cylindrical container that holds liquids
base
baseball bat an implement used in baseball by the batter
baseball glove the handwear used by fielders in playing baseball
basket a container used to hold or carry things, typically made from interwoven strips of cane or wire.
bathtub
bathroom counter a long, flat, narrow surface in a bathroom at waist-height
bear massive plantigrade carnivorous or omnivorous mammals with long shaggy coats and strong claws
bed
bench a long seat for more than one person
bicycle a wheeled vehicle that has two wheels and is moved by foot pedals
bicyclist a person who rides a bicycle
bike rack a rack for parking bicycles
billboard
bird Warm-blooded egg-laying vertebrates characterized by feathers and forelimbs modified as wings
blanket
boat-ship
book
bookshelf
bottle a glass or plastic vessel used for storing drinks or other liquids; typically cylindrical without handles and with a narrow neck that can be plugged or capped
bowl
box
bridge a structure that allows people or vehicles to cross an obstacle such as a river or canal or railway etc.
broccoli plant with dense clusters of tight green flower buds
building a structure that has a roof and walls and stands more or less permanently in one place. Includes houses, grandstands, booths, towers, and skyscrapers
bulletin board
bus a large vehicle carrying many passengers; used for public transport. Typically has swing doors, more than 4 wheels, tires larger than a regular car, and an illuminated front sign
cabinet A case or cupboard usually having doors and shelves. Can be found in a bathroom, kitchen, or other room. Cabinets are not for storing clothes.
cake baked goods made from or based on a mixture of flour, sugar, eggs, and fat
car
carrot deep orange edible root of the cultivated carrot plant
case a glass container used to store and display items in a shop or museum or home
cat feline mammal usually having thick soft fur and no ability to roar: domestic cats
cctv camera camera that can produce images or recordings for surveillance or other private purposes
ceiling
cell phone a phone with access to a cellular radio system so it can be used over a wide area, without a physical connection to a network; a mobile phone.
chair_other Any chair that does NOT fall into the following categories: armchair, stool, seat, sofa. A chair is defined as a seat for one person, with a support for the back.
chandelier branched lighting fixture; often ornate; hangs from the ceiling
chest of drawers furniture with drawers for keeping clothes
clock a timepiece that shows the time of day
column (architecture) a tall vertical cylindrical structure standing upright and used to support a structure
conveyor belt a moving belt that transports people or objects (as in a factory). Includes moving walk-ways and baggage carousels
couch an upholstered seat for more than one person (also known as sofa)
counter-other a long, flat, narrow surface used for making transactions in a store or in a home kitchen (fixed against a wall for preparing food).
cow cattle that are reared for their meat
crib baby bed with high sides made of slats
cup a container for holding liquids while drinking; includes mugs, and could be made of plastic, glass, ceramic, paper, or styrofoam. Cups include drinking glasses without a stem; if the cup/glass has a stem, it is a wine glass
curtain-other hanging cloth used as a blind (especially for a window). does not include shower curtains.
desk A piece of furniture with a flat or sloped surface with drawers, at which one can read,write, or do other work. A table, frame, or case with a sloping or horizontal surface especially for writing and reading

and often with drawers, compartments, and pigeon holes.Includes tables used as desks.

dishwasher a machine for washing dishes
dog a member of the genus Canis (probably descended from the common wolf) that has been domesticated by man since prehistoric times; occurs in many breeds
donut a small ring-shaped friedcake
door a swinging or sliding barrier that will close the entrance to a room or building or vehicle
elephant five-toed pachyderm
escalator a stairway whose steps move continuously on a circulating belt
fan a device for creating a current of air by movement of a surface or surfaces
fence
fire hydrant an upright hydrant for drawing water to use in fighting a fire
fireplace an open recess in a wall at the base of a chimney where a fire can be built
flag emblem usually consisting of a rectangular piece of cloth of distinctive design
floor The inside lower horizontal surface (as of a room, hallway, tent, or other structure). Must be indoors. This includes homogenous floor coverings that span an entire room, such as carpet, linoleum, or marble.
food other any food that is not a banana, apple, sandwich, orange, broccoli, carrot, hot dog, pizza, donut, cake, or fruit
fork cutlery used for serving and eating food
fountain an ornamental structure in a pool or lake from which one or more jets of water are pumped into the air. Includes the jet of water.
frisbee a light, plastic disk propelled with a flip of the wrist for recreation or competition
fruit other any fruit, not an apple, orange, or banana (e.g. pineapple, melon, pear, kiwi, avocado)
giraffe tallest living quadruped; having a spotted coat and small horns and very long neck and legs; of savannahs of tropical Africa
gravel very small rock fragments and pebbles
guard_rail a strong fence/railing at the side of a road or in the middle of an expressway, intended to reduce the risk of serious accidents. A guardrail sits low to the ground and is made of thick horizontal metal rails.
hair_dryer
horse a hand-held electric blower that can blow warm air onto the hair
hot dog Solid-hoofed Herbivorous quadruped domesticated since prehistoric times
junction box A metal box outdoors, containing a junction of electric wires or cables
keyboard a freestanding panel of keys that operate a computer or Typewriter. Should not be part of a laptop (to prevent clashes). Connected with a cord to a computer.
kitchen-island an unattached counter in a kitchen that permits access from all sides
kite plaything consisting of a light frame covered with tissue paper; flown in wind at end of a string. Includes kites for wind-surfing.
knife edge tool used as a cutting instrument; has a pointed blade with a sharp edge and a handle
lamp a device for giving light that has a covering, but allows light to shine through or around. Could be freestanding on a floor and be covered with a shade, or could hang from the ceiling, or be placed on a table/desk/nightstand.
laptop a portable computer small enough to use in your lap. Includes the laptop keyboard.
light-other Any other type of light, e.g. fluorescent ceiling lights, bathroom lights embedded into ceiling, etc
mailbox a private box for delivery of mail
microwave kitchen appliance that cooks food by passing an electromagnetic wave through it
mirror polished surface that forms images by reflecting light
motorcycle a motor vehicle with two wheels and a strong frame
motorcyclist a human relying on a motorcycle/moped for movement. Must be actively riding the motorcycle/moped (not standing nearby it).
mountain hill
mouse a computer input device that controls an on-screen pointer
net an open fabric of string or rope or wire woven together at regular intervals
night stand a small low bedside table, typically having drawers. a small bedside table or stand
orange round yellow to orange fruit of any of several citrus trees
ottoman a low upholstered seat or thick cushion used as a seat
oven
painting graphic art consisting of an artistic composition made by applying paints to a surface
paper a material made of cellulose pulp derived mainly from wood or rags or certain grasses. Includes loose paper, reams of paper, paper towels, and napkins
parking meter a money-operated timer located next to a parking space; depositing money into it entitles you to park your car therefore a specified length of time
person_nonrider a human being who is not a motorcyclist, bicyclist, or rider other (could be a motor vehicle passenger)
pier_wharf Either a pier or a wharf. A pier is a platform/low structure built out from the shore into the water and supported by piles; provides access to ships and boats. A wharf is a level quayside area to which a ship may be moored to load and unload at the edge of water.
pillow
pizza Italian open pie made of thin bread dough spread with a spiced mixture of e.g. tomato sauce and cheese
plate
platform area alongside a railway track providing convenient access to trains, or a ramp/raised surface at a skatepark
playingfield a field used for outdoor team games. Includes tennis courts, baseball fields/diamonds, and fields for soccer, ultimate frisbee, and equestrian events.
plaything_other a child’s toy, or something used like a toy, that is not a teddy bear.
pole a long (usually round) rod of wood or metal or plastic. includes lamp posts (excluding the light) and streetlight poles (excluding the light)
pool table game equipment consisting of a heavy table on which pool is played
poster
potted plant a plant that is planted and grown in a container rather than in the ground
radiator heater consisting of a series of pipes for circulating steam or hot water to heat rooms or buildings
railing-banister a barrier consisting of a horizontal bar and supports; a railing at the side of a staircase or balcony to prevent people from falling
railroad a line of track providing a runway for wheels
range_hood exhaust hood over a kitchen range
refrigerator a refrigerator in which the coolant is pumped around by an electric motor
remote a device that can be used to control a machine or apparatus from a distance
rider_other a human relying upon a segway, skateboards, electric scooter, lawn mower, rickshaw, wheelchair etc (any other device that is not a motorcycle, bicycle, or vehicle)for movement on land.
river_lake River or lake. River: a large natural stream of water, with possible rapids. Lake: a body of (usually fresh) water

surrounded by land. Lakes generally have very still water. Should appear naturally formed.

road a long, narrow stretch with a smoothed or paved surface, made for traveling by motor vehicle, between two or more points; street or highway. Often bounded by curbs.
road_barrier concrete safety barriers on roads and highways; also known as Jersey barrier
rock a lump or mass of hard consolidated mineral matter
rug_floormat a floor covering of thick woven material or animal skin– just with limited size, covers only a small section of the floor. Includes rugs, and floormats (door mats, bath mat). Does not include carpet.
runway a strip of level paved surface where planes can take off,land, or taxi from a terminal;or a narrow platform extending from the stage into the audience in a theater or nightclub etc.
sandwich two (or more) slices of bread with a filling between them
scissors an edge tool having two crossed pivoting blades
sconce A decorative wall bracket/object for holding light bulbs, candles or other sources of light. Must be attached to the side of a wall.
sculpture statue or a three-dimensional work of art
sea A sea or ocean. Often has large amounts of waves/surf and a sandy beach. May have surfers on surfboards. Unlike a river or a lake, one generally cannot see the far bank of a sea/ocean.
seat a space reserved for sitting (as in a theater, auditorium, stadium, on a train or airplane, or in a car). (that is not a bench,chair, stool, sofa, or armchair)
sheep woolly usually horned ruminant mammal related to the goat
shelf
shower
shower_curtain fabric found next to a bathtub or shower to keep water from spilling out into the bathroom
sidewalk_pavement walk consisting of a paved area for pedestrians; usually beside a street or roadway (not a road)
sink plumbing fixture consisting of a water basin fixed to a wall or floor and having a drain-pipe (usually in a bathroom or kitchen)
skateboard a board with wheels that is ridden in a standing or crouching position and propelled by foot
skis sports equipment for skiing on snow
sky the atmosphere and outer space as viewed from the earth
slow wheeled object any slow-moving wheeled object, such as strollers, wheelchairs, and carts
snow a layer of snowflakes (white crystals of frozen water) covering the ground
snowboard a board that resembles a broad ski or a small surfboard; used in a standing position to slide down snow-covered slopes
spoon a piece of cutlery with a shallow bowl-shaped container and a handle; used to stir or serve or take up food
sports ball round object that is hit or thrown or kicked in games
stage a large platform on which people can stand and can be seen by an audience
stairs
stool a simple seat without a back or arms; or a high seat at a bar without arms
storage_tank a large (usually metallic) vessel for holding gases or liquids
stove
streetlight a lamp or light supported on a lamppost/pole; for illuminating a street
suitcase a portable rectangular container for carrying clothes
surfboard a narrow buoyant board for riding surf
swimming_pool pool that provides a facility for swimming
swivel_chair a chair that swivels on its base. Can be turned around a central point to face in a different direction without moving the legs
table A piece of furniture consisting of a smooth flat slab fixed on legs. A piece of furniture with a flat top and one or more legs, providing a level surface on which objects may be placed, and that can be used for such purposes as eating, writing, working, or playing games.
teddy_bear Stuffed toy in the form of a bear. It is a type of plaything/child’s toy (usually plush and stuffed with soft materials).
television
tennis racket a racket used to play tennis
tent a portable shelter (usually of canvas stretched over supporting poles and fastened to the ground with ropes and pegs). For example, a camping tent.
terrain dirt, sand, grass, or any kind of horizontally spreading vegetation
tie neckwear consisting of a long narrow piece of material worn (mostly by men) under a collar and tied in knot at the front
toaster
toilet a plumbing fixture for defecation and urination
toothbrush small brush; has long handle;used to clean teeth
towel a thick, rectangular piece of absorbent cloth for drying or wiping
traffic light a visual signal to control the flow of traffic at intersections (includes stoplights, pedestrian crossing lights, etc)
traffic sign a sign usually on the side of a street or highway bearing symbols or words of warning or direction to motorists or pedestrians
trailer
train All vehicles that move on rails,

e.g. trams, trains.

trash can a bin that holds rubbish until it is collected
tray a flat, shallow container with a raised rim, typically used for carrying food and drink, or for holding small items.
truck
tunnel a passageway through or under something, usually underground (especially one for trains or cars)
umbrella a lightweight collapsible canopy. Could be handheld or mounted above a table outdoors.
vase an open jar of glass, porcelain,or metal used as an ornament or to hold flowers
vegetation Trees, hedges, and all kinds of vertically growing vegetation (for example, must be higher than lawn grass)
wall Either: one of the sides of a room or building connecting the floor and ceiling. Or: a masonry fence around a garden,park, or estate.
wardrobe a tall piece of furniture that provides storage space for clothes; has a door and rails or hooks for hanging clothes
washer-dryer a home appliance for washing or drying clothes and linens automatically
waterfall a steep descent of the water of a river
water_other all other kinds of water, e.g. puddles, ditchwater, aquariums, shallow artificially made ponds, zoo ponds, flooded areas, bathtub water, water coming out of a fire hydrant, water from a kitchen faucet, etc.is not a lake, river, fountain,swimming pool, or sea.
whiteboard a hard smooth white surface used for writing or drawing on with markers.
window a framework of wood or metal that contains a glass window-pane and is built into a wall or roof to admit light or air
window-blind A blind for privacy or to keep out light. Placed indoors on a window.
wine_glass a glass that has a stem and in which wine is served
zebra any of several fleet black-and-white striped African equines

Code/Model/Publication URL : MSeg: A Composite Dataset for Multi-domain Semantic Segmentation

Source URL: MSeg

Dataset Examples :

Credits : MSeg: A Composite Dataset for Multi-domain Semantic Segmentation

26. PASCAL-Context Dataset

      Dataset Characteristics :

Task Type: semantic segmentation, object segmentation

Image Count : 10,103

Label Count : 4,203

Categories : 540

Code/Model/Publication URL: The Role of Context for Object Detection and Semantic Segmentation in the Wild

Source URL: PASCAL-Context Dataset

Dataset Examples :

Credits : PASCAL-Context Dataset

 

27. LVIS: A DATASET FOR LARGE VOCABULARY INSTANCE SEGMENTATION

     Dataset Characteristics :

Task Type: Instance segmentation

Image Count: ~160k images

Label Count: ~2M instance annotations

Categories: 1203

Code/Model/Publication URL : LVIS: A Dataset for Large Vocabulary Instance Segmentation

Source URL: LVIS Dataset

Dataset Examples :

Credits: LVIS Dataset

 

28. Flower Datasets

      Dataset Characteristics :

Task Type: flower segmentation

Image Count : Dataset 1 : 1360 images consisting of 17 flower species

                          Dataset 2 : 8,189 images consisting of 102 flower species

Categories : 119 categories

Dataset 1 :

Flowers Class
Buttercup
Colts’ Foot
Daffodil
Dandelion
Daisy
fritillary
Iris
Pansy
Sunflower
Windflower
Snowdrop
Lily Valley
Bluebell
Crocus
Tigerlily
Tulip
Cowslip

Dataset 2 :

Category #images Category #images Category #images
alpine sea holly alpine sea holly 43 buttercup buttercup 71 fire lily fire lily 40
anthurium anthurium 105 californian poppy californian poppy 102 foxglove foxglove 162
artichoke artichoke 78 camellia camellia 91 frangipani frangipani 166
azalea azalea 96 canna lily canna lily 82 fritillary fritillary 91
ball moss ball moss 46 canterbury bells canterbury bells 40 garden phlox garden phlox 45
balloon flower balloon flower 49 cape flower cape flower 108 gaura gaura 67
barbeton daisy barbeton daisy 127 carnation carnation 52 gazania gazania 78
bearded iris bearded iris 54 cautleya spicata cautleya spicata 50 geranium geranium 114
bee balm bee balm 66 clematis clematis 112 giant white arum lily giant white arum lily 56
bird of paradise bird of paradise 85 colt's foot colt’s foot 87 globe thistle globe thistle 45
bishop of llandaff bishop of llandaff 109 columbine columbine 86 globe-flower globe-flower 41
black-eyed susan black-eyed susan 54 common dandelion common dandelion 92 grape hyacinth grape hyacinth 41
blackberry lily blackberry lily 48 corn poppy corn poppy 41 great masterwort great masterwort 56
blanket flower blanket flower 49 cyclamen cyclamen 154 hard-leaved pocket orchid hard-leaved pocket orchid 60
bolero deep blue bolero deep blue 40 daffodil daffodil 59 hibiscus hibiscus 131
bougainvillea bougainvillea 128 desert-rose desert-rose 63 hippeastrum hippeastrum 76
bromelia bromelia 63 english marigold english marigold 65 japanese anemone japanese anemone 55
king protea king protea 49 peruvian lily peruvian lily 82 stemless gentian stemless gentian 66
lenten rose lenten rose 67 petunia petunia 258 sunflower sunflower 61
lotus lotus 137 pincushion flower pincushion flower 59 sweet pea sweet pea 56
love in the mist love in the mist 46 pink primrose pink primrose 40 sweet william sweet william 85
magnolia magnolia 63 pink-yellow dahlia? pink-yellow dahlia? 109 sword lily sword lily 130
mallow mallow 66 poinsettia poinsettia 93 thorn apple thorn apple 120
marigold marigold 67 primula primula 93 tiger lily tiger lily 45
mexican aster mexican aster 40 prince of wales feathers prince of wales feathers 40 toad lily toad lily 41
mexican petunia mexican petunia 82 purple coneflower purple coneflower 85 tree mallow tree mallow 58
monkshood monkshood 46 red ginger red ginger 42 tree poppy tree poppy 62
moon orchid moon orchid 40 rose rose 171 trumpet creeper trumpet creeper 58
morning glory morning glory 107 ruby-lipped cattleya ruby-lipped cattleya 75 wallflower wallflower 196
orange dahlia orange dahlia 67 siam tulip siam tulip 41 water lily water lily 194
osteospermum osteospermum 61 silverbush silverbush 52 watercress watercress 184
oxeye daisy oxeye daisy 49 snapdragon snapdragon 87 wild pansy wild pansy 85
passion flower passion flower 251 spear thistle spear thistle 48 windflower

windflower 54
pelargonium pelargonium 71 spring crocus spring crocus 42 yellow iris yellow iris 49

Credits : 102 Category Flower Dataset

Code/Model/Publication URL: Automated flower classification over a large number of classes

Source URLFlower Dataset

Dataset Examples :

Credits: Automated flower classification over a large number of classes

 29. NYU Depth Dataset V2

       Dataset Characteristics :

Task Type: Dense Semantic Image Segmentation, Indoor segmentation

Image Attribute: 1449 densely labeled pairs of aligned RGB and depth images

                                    464 different indoor scenes across 26 scene classes

            5,064 distinct objects,spanning 894 different classes

Code/Model/Publication URL: Indoor Segmentation and Support Inference from RGBD Images

Source URL: NYU Depth Dataset V2

Dataset Examples :

Credits: NYU Depth Dataset V2

 

30. Fashionista

     Dataset Characteristics :

Task Type: Clothing Estimation

Image Attribute : 158,235 fashion photos

Categories :

Garments

background(null) top  cardigan
skin skirt blazer
hair jacket t-shirt
dress coat socks
bag shirt necklace
blouse shoes bracelet
purse accessories boots
sweater jumper cape
leggings romper vest
belt tights jeans
heels wedges stockings
hat shorts

Code/Model/Publication URL: Parsing Clothing in Fashion Photographs

Source URL: Clothing Parsing

Dataset Examples :

Credits : Parsing Clothing in Fashion Photographs

 

31. UNIMIB Food Database

      Dataset Characteristics :

Task Type: food segmentation

Image Attribute: 1,027 canteen trays for a total of 3,616 food instances belonging to 73 food classes.

Segmented images of the 73 food categories

Code/Model/Publication URL: Food recognition: a new dataset, experiments, and results

Source URL: UNIMIB Food Database

Dataset Examples :

Credits: Food recognition: a new dataset, experiments, and results

 

32. SYNSCAPES

      Dataset Characteristics :

Task Type: Semantic segmentation

Image Attribute: 25,000 RGB images in PNG format at 1440×720 resolution

Categories: 19

Classes

Road SideWalk
Building Wall
Fence Pole
Tr.Sign Tr. Light
Vegetation Terrain
Sky Person
Rider Car
Truck Bus
Train Motorcycle
Bicycle

Code/Model/Publication URL : Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing

Source URL: SYNSCAPES

Dataset Examples :

Credits : Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing

 

33. YouTube-VOS

       Dataset Characteristics :

Task Type :

    • Semi-supervised Video Object Segmentation
    • Video Instance Segmentation

Video Attribute

    • 4000+ high-resolution YouTube videos
    • 90+ semantic categories
    • 7800+ unique objects
    • 190k+ high-quality manual annotations
    • 340+ minutes duration

Object categories in YouTube-VOS

person cat train hedgehog squirrel
table ape snake owl eagle
rope camera parrot zebra plant
snail chameleon watch giant_panda giraffe
airplane toilet box stuffed_toy tissue
sedan bear bus camel guitar
lizard fox shark frisbee kangaroo
microphone duck leopard tiger whale
cloth cup dog elephant surfboard
knife bottle shovel skateboard horse
earless_seal tennis_racket small_panda flag monkey
others frog crocodile spider mirror
sheep deer mouse umbrella ball
ring fish motorbike boat paddle
jellyfish necklace rabbit turtle snowboard
raccoon eyeglasses ant hat bird
penguin parachute backpack cow truck
lion bucket butterfly hand dolphin
sign bike handbag

Code/Model/Publication URL :

Source URL : YouTube-VOS

Dataset Examples :

Credits : Video Instance Segmentation

 

34. The Lane Marker Dataset

       Dataset Characteristics :

Task Type : binary marker segmentation, lane-dependent pixel-level segmentation

Image Attribute100,042 labeled lane marker images from about 350 km recorded drives

Code/Model/Publication URL : Unsupervised Labeled Lane Markers Using Maps

Source URL : The Lane Marker Dataset

Dataset Examples :

Credits : Unsupervised Labeled Lane Markers Using Maps

 

35. Total-Text

      Dataset Characteristics :

Task Type : Scene Text Detection, Segmentation-based text detection

Image Attribute : It has 1,555 scene images, 4,265 curved text, 9,330 annotated words with 3 different text orientations including horizontal, multi-oriented, and curved text.

Code/Model/Publication URL : Total-Text: A Comprehensive Dataset for Scene Text Detection and Recognition

Source URL : The Lane Marker Dataset

Dataset Examples :

Credits : Total-Text: A Comprehensive Dataset for Scene Text Detection and Recognition

 

36. MVTec D2S: Densely Segmented Supermarket Dataset

      Dataset Characteristics :

Task Type : instance-aware semantic segmentation in an industrial domain

Image Attribute : 21,000 images in 700 different scenes with various backgrounds, clutter objects, and occlusion levels.

Categories : 60

adelholzener_alpenquelle_classic_075carrot
adelholzener_classic_naturell_02grapes_green_sugraone_seedless
apple_roter_boskoopadelholzener_classic_bio_apfelschorle_02
banana_singlesalad_iceberg
caona_kakaohaltiges_getraenkepulverbanana_bundle
clementinelettuce
clementine_singleavocado
coca_cola_light_05apple_golden_delicious
corny_nussvoll_singleaugustiner_lagerbraeu_hell_05
douwe_egberts_professional_kaffee_gemahalencorny_schoko_banana_single
ethiquable_gruener_tee_ceyloncocoba_fruehstueckskakao_mit_honing
franken_tafelreinigaradelholzener_gourmet_mineralwasser_02
gepa_bio_und_fair_fenchelteegrapes_sweet_celebration_seedless
gepa_bio_und_fair_pfefferminzteecoca_cola_05
koelln_muesli_schokotegernseer_hell_03
pasta_reggia_fusillikoelln_muesli_fruechte
pasta_reggia_spaghettifeldsalat
rispentomatengepa_bio_und_fair_kamillentee
roma_rispentomatendr_oetker_vitalis_knuspermuesli_klassisch
suntory_gokuri_limonadecucumber
pearorange_single
kilimanjaro_tea_earl_greykiwi
zucchiniapple_granny_smith
pelikan_tintenpatrone_canongepa_bio_caffe_crema
cafe_wunderbar_espressogepa_italienischer_bio_espresso
gepa_bio_und_fair_rooibosteecorny_nussvoll
augustiner_weissbier_05pasta_reggia_elicoidali
corny_schoko_bananarucola
adelholzener_alpenquelle_naturell_075apple_braeburn_bundle
gepa_bio_und_fair_kraeuterteemischungoranges

Code/Model/Publication URLMVTec D2S: Densely Segmented Supermarket Dataset

Source URL : MVTec D2S Dataset

Dataset Examples :

Credits : MVTec D2S: Densely Segmented Supermarket Dataset

 

37. CORe50

      Dataset Characteristics :

Task Type : image segmentation

Video Attribute :   50 domestic objects videos collected in 11 distinct sessions (8 indoor and 3 outdoor) characterized by different backgrounds and lighting.

Categories : 10

plug adapters
mobile phones
scissors
light bulbs
cans
glasses
balls
markers
cups
remote controls

Code/Model/Publication URL : CORe50: a New Dataset and Benchmark for Continuous Object Recognition

Source URL : CORe50

Dataset Examples :

Credits : CORe50: a New Dataset and Benchmark for Continuous Object Recognition

 

38. Fashionpedia

     Dataset Characteristics :

Task Type : Apparel instance segmentation

Image Attribute : 48,825 clothing images in daily-life, street-style,celebrity events, runway, and online shopping with 46 apparel categories and 294 attributes

Code/Model/Publication URL : Fashionpedia: Ontology, Segmentation, and an Attribute Localization Dataset

Source URL : Fashionpedia

Dataset Examples :

Credits : Fashionpedia: Ontology, Segmentation, and an Attribute Localization Dataset

 

39. The Oxford-IIIT Pet Dataset

      Dataset Characteristics :

Task Type : image segmentation

Image Attribute : 7349 images of cats and dogs

Categories :

Code/Model/Publication URL : Cats and Dogs

Source URL : The Oxford-IIIT Pet Dataset

Dataset Examples :

Credits : The Oxford-IIIT Pet Dataset

 

40. BDD100K: A Large-scale Diverse Driving Video Database

      Dataset Characteristics :

Task Type : drivable area segmentation, semantic segmentation, instance segmentation, multi-object segmentation tracking

Video Attribute : 100K videos and 10 tasks

Code/Model/Publication URL : BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning

Source URL : BDD100K

Dataset Examples :

Credits : BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning

 

41. MOTS Dataset

      Dataset Characteristics :

Task Type : Multi-Object Tracking and Segmentation

Video Attribute : 8 challenging video sequences (4 training, 4 test) in unconstrained environments filmed with both static and moving cameras

Code/Model/Publication URL : MOTS: Multi-Object Tracking and Segmentation

Source URL : MOTS

Dataset Examples :

Credits : MOTS: Multi-Object Tracking and Segmentation

 

42. KITTI MOTS Dataset

      Dataset Characteristics :

Task Type : Multi-Object Tracking and Segmentation

Video Attribute : 21 training sequences and 29 test sequences

Code/Model/Publication URL : MOTS: Multi-Object Tracking and Segmentation

Source URL : KITTI-MOTS

Dataset Examples :

Credits : MOTS: Multi-Object Tracking and Segmentation

 

43. ApolloScape Dataset

      Dataset Characteristics :

Task Type Lanemark segmentation

Image Attribute : 100K image frames, 80k lidar point cloud and 1000km trajectories for urban traffic

Code/Model/Publication URL : DVI: Depth Guided Video Inpainting for Autonomous Driving

Source URL : ApolloScape

Dataset Examples :

Credits : Toolkit for ApolloScape Dataset

 

44. Camouflaged Object (CAMO) dataset

      Dataset Characteristics :

Task Type : camouflaged object segmentation

Image Attribute : 1250 images under two categories, i.e., naturally camouflaged objects and artificially camouflaged objects.

Categories : Camouflaged animals consist of amphibians, birds, insects, mammals, reptiles, and underwater animals in various environments, i.e., ground, underwater, desert, forest, mountain, and snow.

Camouflaged human falls into soldiers on the battlefields and human body painting arts.

Code/Model/Publication URL : Anabranch Network for Camouflaged Object Segmentation

Source URL : Camouflaged Object Segmentation

Dataset Examples :

Credits : Anabranch Network for Camouflaged Object Segmentation

 

45. Berkeley Segmentation Data Set and Benchmarks 500 (BSDS500)

      Dataset Characteristics :

Task Type : image segmentation

Image Attribute :  300 images are used for training / validation and 200 fresh images, together with human annotations, are added for testing

Code/Model/Publication URL : Contour Detection and Hierarchical Image Segmentation

Source URL : Berkeley Segmentation Data Set and Benchmarks 500 (BSDS500)

Dataset Examples :

Credits : Contour Detection and Hierarchical Image Segmentation

 

46. Microsoft Research Cambridge Object Recognition Image Database (MSRC v1 & v2)

      Dataset Characteristics :

Task Type : image segmentation, semantic segmentation

Image Attribute :  v1 : 240 images and 9 object classes with coarse pixel-wise labeled images.

  v2 : 591 images, 23 object classes with accurate pixel-wise labeled images.

Code/Model/Publication URL : Microsoft Research Cambridge Object Recognition Image Database

Source URL : Microsoft Research Cambridge Object Recognition Image Database

Dataset Examples :

Credits : Microsoft Research Cambridge Object Recognition Image Database

 

47. Transmission Electron Microscopic (TEM) Cell Recordings Segmentation Dataset

      Dataset Characteristics :

Task Type : automatic segmentations for mitochondria

Image Attribute :  A minimum of 10000 cells were recorded under five classes (background, cytoplasm,nucleus, mitochondria and vesicles)

Code/Model/Publication URL : Semi-automatic procedure for the determination of the cell surface area used in systems immunology

Source URL : Transmission Electron Microscopic (TEM) Cell Recordings Segmentation Dataset

Dataset Examples :

Credits : Semi-automatic procedure for the determination of the cell surface area used in systems immunology

 

48. Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59)

      Dataset Characteristics :

Task Type : Motion Segmentation

Video Attribute :  59 video sequence with 720 frames annotated

Code/Model/Publication URL : Segmentation of moving objects by long term video analysis

Source URL : Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59)

Dataset Examples :

Credits : Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59)

 

49. Georgia Tech Segmentation and Tracking Dataset

      Dataset Characteristics :

Task Type target segmentation and tracking in video

Video Attribute : 6 videos, 243 frames

parachutegirl
monkeydogpenguin
birdfallcheetah

Code/Model/Publication URL : Motion Coherent Tracking Using Multi-label MRF Optimization

Source URL : GaTech SegTrack

Dataset Examples :

Credits : GaTech SegTrack

 

50. Video Segmentation Benchmark (VSB100) Dataset

      Dataset Characteristics :

Task Type : Video Segmentation

Video Attribute :  100 HD quality videos with ground truth annotations

Code/Model/Publication URL : A Unified Video Segmentation Benchmark: Annotation, Metrics and Analysis

Source URL : Video Segmentation Benchmark (VSB100) Dataset

Dataset Examples :

Credits : A Unified Video Segmentation Benchmark: Annotation, Metrics and Analysis

 

51. Scene Flow Datasets

      Dataset Characteristics :

Task Type Object-level and material-level segmentation

Image Attribute :  more than 39000 stereo frames in 960×540 pixel resolution

Code/Model/Publication URL : A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation

Source URL : Scene Flow Datasets

Dataset Examples :

Credits : A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation

 

52. FreiHAND Dataset

      Dataset Characteristics :

Task Type hand pose and shape estimation from single color image

Image Attribute :  130240 training and 3960 evaluation samples with Hand segmentation mask (224×224 pixels)

Code/Model/Publication URL : FreiHAND: A Dataset for Markerless Capture of Hand Pose and Shape from Single RGB Images

Source URL : FreiHAND Dataset

Dataset Examples :

Credits : FreiHAND: A Dataset for Markerless Capture of Hand Pose and Shape from Single RGB Images

 

53. YouTube-Objects Dataset

      Dataset Characteristics :

Task Type video segmentation,  foreground object segmentation

Video Attribute :  26 web videos with 10 object classes and more than 20,000 frames

aeroplanebird
boat car
cat cow
dog horse
motorbiketrain

Code/Model/Publication URL : Supervoxel-Consistent Foreground Propagation in Video

Source URL : YouTube-Objects Segmentation Labels

Dataset Examples :

Credits : Supervoxel-Consistent Foreground Propagation in Video

 

54. SegTrack v2 Dataset

      Dataset Characteristics :

Task Type video segmentation

Video Attribute :  1,000 frames with pixel-level annotations for 14 categories

GirlBirdfall
ParachuteCheetah
MonkeydogPenguin
Drifting CarHummingbird
FrogWorm
SoldierMonkey
Bird of ParadiseBMX

Code/Model/Publication URL : Video Segmentation by Tracking Many Figure-Ground Segments

Source URL : SegTrack v2 Dataset

Dataset Examples :

Credits : SegTrack v2 Dataset

 

55. Human semantic part segmentation Dataset

      Dataset Characteristics :

Task Type ground and aerial robot segmentation

Image Attribute :  two hundred and one images of six different people in multiple viewpoints

Categories :  14 body parts

HeadTorso
Left Upper armLeft Lower arm
Left handRight Upper hand
Right Lower armRight hand
Right Upper legRight Lower leg
Right footLeft Upper leg
Left Lower legLeft foot

Code/Model/Publication URL : Deep Learning for Human Part Discovery in Images

Source URL : Human semantic part segmentation Dataset

Dataset Examples :

Credits : Deep Learning for Human Part Discovery in Images

 

56. Cambridge-driving Labeled Video Database (CamVid)

      Dataset Characteristics :

Task Type : semantic segmentation

Image Attribute :  per-pixel semantic segmentation of over 700 images

Categories :  32 semantic classes

List of the 32 object class names and their corresponding colors used for labeling.

Code/Model/Publication URL : Semantic Object Classes in Video: A High-Definition Ground Truth Database

Source URL : Human semantic part segmentation Dataset

Dataset Examples :

Credits : Semantic Object Classes in Video: A High-Definition Ground Truth Database

 

57. Cityscapes Dataset

      Dataset Characteristics :

Task Type instance segmentation, semantic segmentation

Image Attribute : 25k Images

Group

Classes

flat road, sidewalk, parking, rail track
human person, rider
vehicle car, truck, bus, on rails, motorcycle, bicycle, caravan, trailer
construction building, wall, fence, guard rail, bridge, tunnel
object pole, pole group,  traffic sign, traffic light
nature vegetation, terrain
sky sky
void ground, dynamic, static

*require institutional login

Code/Model/Publication URLSemantic Understanding of Urban Street Scenes

Source URL : Cityscapes Dataset

Dataset Examples :

Credits : Cityscapes Dataset

 

58. Multi-Human Parsing (MHP) v1.0 Dataset

      Dataset Characteristics :

Task Type : multi-human parsing

Image Attribute : 4,980 images consist of  14,969 person instances with fine-grained annotations at pixel-level

Categories : 18 different semantic labels

Face Hair
Upper ClothesRight Arm
Left ArmPants
Left ShoeRight Shoe
Right LegLeft Leg
Torso SkinDress
HatSkirt
BagSun Glasses
BeltScarf

Code/Model/Publication URL : Multi-Human Parsing in the Wild

Source URL : Multi-Human Parsing (MHP) v1.0 Dataset

Dataset Examples :

Credits : Multi-Human Parsing in the Wild

 

59. Mut1ny head/face segmentation dataset

      Dataset Characteristics :

Task Type : head/face segmentation

Image Attribute : 16.5k (16557) fully pixel-level labeled segmentation images

Categories : 11 facial labels

LipsEyes
NoseHair
EarsEyebrows
TeethGeneral face
Facial hair Specs/sunglasses
Background/undefined

Source URL : Mut1ny head/face segmentation dataset

Dataset Examples :

Credits : Mut1ny head/face segmentation dataset

 

60. Semantic Drone Dataset

      Dataset Characteristics :

Task Type semantic understanding of urban scenes for increasing the safety of autonomous drone flight and landing procedures

Image Attribute : The training set contains 400 publicly available images and the test set is made up of 200 private images.

Semantic classes of the Drone Dataset

  • tree
  • grass
  • other vegetation
  • dirt
  • gravel
  • rocks
  • water
  • paved area
  • pool
  • person
  • dog
  • car
  • bicycle
  • roof
  • wall
  • fence
  • fence-pole
  • window
  • door
  • obstacle

Source URL : Semantic Drone Dataset

Dataset Examples :

Credits : Semantic Drone Dataset

 

61. The SYNTHIA dataset

      Dataset Characteristics :

Task Type : semantic segmentation

Image Attribute : +200,000 HD images from video streams and +20,000 HD images from independent snapshots

Categories : pixel-level semantic annotations for 13 classes

skybuilding
roadsidewalk
fencevegetation
lane-markingpole
cartraffic signs
pedestrianscyclists
miscellaneous

Code/Model/Publication URL : The SYNTHIA Dataset: A Large Collection of Synthetic Images for Semantic Segmentation of Urban Scenes

Source URL : The SYNTHIA dataset

Dataset Examples :

Credits : The SYNTHIA Dataset: A Large Collection of Synthetic Images for Semantic Segmentation of Urban Scenes

 

62. RailSem19

      Dataset Characteristics :

Task Type A Dataset for Semantic Rail Scene Understanding

Image Attribute : 8500 unique images having 110000 annotation taken from a the ego-perspective of a rail vehicle. Over 1000 examples with railway crossings and 1200 tram scenes.

Categories : pixel-level semantic annotations for 13 classes

buffer-stopcrossing
guardrailtrain-car
platformrail
switch-ind.switch-left
switch-rightswitch-unknown
switch-statictrack-sign-front
track-signal-backtrack-signal-front
person-grouptruck
carfence
personpole
rail-occluder

Code/Model/Publication URL : RailSem19: A Dataset for Semantic Rail Scene Understanding

Source URL : RailSem19

Dataset Examples :

Credits : RailSem19: A Dataset for Semantic Rail Scene Understanding

 

63.  DeepFashion: In-shop Clothes Retrieval Dataset

      Dataset Characteristics :

Task Type : instance semantic segmentation

Image Attribute : 7,982 number of clothing items, 52,712 number of in-shop clothes images, and ~200,000 cross-pose/scale pairs;

Code/Model/Publication URL : DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations

Source URL : DeepFashion: In-shop Clothes Retrieval

Dataset Examples :

Credits : DeepFashion: In-shop Clothes Retrieval

 

64. Indian Driving Dataset

      Dataset Characteristics :

Task Type : road scene understanding in unstructured environments

Image Attribute : 10,004 images, finely annotated with 34 classes collected from 182 drive sequences on Indian roads

roadparking
drivable fallbackground
sidewalkrail track
non-drivable fallbacktrain
personanimal
ridermotorcycle
bicycleautorickshaw
cartruck
buscaravan
trailertrain
vehicle fallbackcurb
wallfence
guard railbillboard
traffic signtraffic light
polepolegroup
obs-str-bar-fallbackbuilding
bridgetunnel
vegetationsky
fallback background

Code/Model/Publication URL : IDD: A Dataset for Exploring Problems of Autonomous Navigation in Unconstrained Environments

Source URL : Indian Driving Dataset

Dataset Examples :

Credits : IDD: A Dataset for Exploring Problems of Autonomous Navigation in Unconstrained Environments

 

65. A2D2 Dataset

      Dataset Characteristics :

Task Type : Instance Segmentation, Semantic Segmentation

Image Attribute : 41,277 camera images are semantically labelled.

Natural ObjectSky
RD normal streetBuilding
CarSideWalk
Ego carTruck
Grid StructureRoad Blocks
Drivable cobblestoneCurbstone
Solid lineNon-drivable street
PolesDashed lines
Irrelevant SignsTraffic sign
Parking areaObstacles/trash
RD restricted areaTraffic guide obj
Slow Drive areaSignal corpus
PedestrianBicycle
Painted driv. instrutility vehicle
traffic signalSidebars
Small vehicleszebra crossing
Electronic trafficTractor
Blurred AreaAnimals
Speed bumperRain dirt

Code/Model/Publication URL : A2D2: Audi Autonomous Driving Dataset

Source URL : A2D2 Dataset

Dataset Examples :

Credits : A2D2: Audi Autonomous Driving Dataset

 

66. Mapillary Vistas Dataset 

      Dataset Characteristics :

Task Type : semantic image segmentation and instance-specific image segmentation

Image Attribute : 25 000 high-resolution images annotated into 66 object categories with additional, instance-specific labels for 37 classes.

BirdGround Animal
Crosswalk-PlainPerson
BicyclistMotorcyclist
other riderLane Marking - Crosswalk
BannerBench
Bike RackBillboard
Catch BasinCCTV Camera
Fire HydrantJunction Box
MailboxManhole
Phone BoothStreet Light
PoleTraffic Sign Frame
Utility PoleTraffic Light
Traffic Sign (Back)Traffic Sign (Front)
Trash CanBicycle
BoatBus
CarCaravan
MotorcycleOther Vehicle
TrailerTruck
Wheeled Slow

*require institutional login

Code/Model/Publication URL : The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes

Source URL : Mapillary Vistas Dataset 

Dataset Examples :

Credits : The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes

 

67. eTRIMS Image Database

      Dataset Characteristics :

Task Type object segmentation and class segmentation of real world street scene.

Image Attribute : The database is comprised of two datasets,

The 4-Class eTRIMS Dataset with 4 annotated object classes with 60 images and 534 annotated objects.

  • sky
  • building
  • vegetation
  • pavement/road

The 8-Class eTRIMS Dataset with 8 annotated object classes with 60 images and 1702 annotated objects.

  • sky
  • building
  • vegetation
  • pavement
  • window
  • door
  • car
  • road

Code/Model/Publication URL : eTRIMS Image Database for Interpreting Images of Man-Made Scenes

Source URL : eTRIMS Image Database

Dataset Examples :

Credits : eTRIMS Image Database

 

68. Daimler Urban Segmentation Dataset

      Dataset Characteristics :

Task Type image segmentation of highly cluttered urban traffic scenes

Image Attribute : 5000 rectified stereo image pairs with a resolution of 1024×440 with pixel-level semantic class annotations into 5 classes: ground, building, vehicle, pedestrian, sky

Code/Model/Publication URL : Efficient Multi-Cue Scene Segmentation

Source URL : Daimler Urban Segmentation Dataset

Dataset Examples :

Credits : Stixmantics: A Medium-Level Model for Real-Time Semantic Scene Understanding

 

69. KITTI Motion Segmentation Dataset

      Dataset Characteristics :

Task Type Motion Segmentation

Image Attributesix sequences from KITTI raw data to generate a total of 1750 frames. In addition to these frames, 200 frames from KITTI scene flow are used to provide us with 1950 frames in total.

Code/Model/Publication URL : MODNet: Moving Object Detection Network with Motion and Appearance for Autonomous Driving

Source URL : KITTI Motion Segmentation Dataset

Dataset Examples :

Credits : MODNet: Moving Object Detection Network with Motion and Appearance for Autonomous Driving

 

70. CUHK DeepFashion2 Dataset

      Dataset Characteristics :

Task Type Cloth Segmentation

Image Attribute : 491K diverse images of 13 popular clothing categories, 801K clothing clothing items, 873K Commercial-Consumer clothes pairs

short sleeve toplong sleeve top
short sleeve outwearlong sleeve outwear
vestsling
shortstrousers
skirtshort sleeve dress
long sleeve dressvest dress
sling dress

Code/Model/Publication URL : DeepFashion2: A Versatile Benchmark for Detection, Pose Estimation,Segmentation and Re-Identification of Clothing Images

Source URL : DeepFashion2

Dataset Examples :

Credits : DeepFashion2

 

71. WildDash 2 Dataset

      Dataset Characteristics :

Task Type : semantic and instance segmentation for the automotive domain

Image Attribute : 4256 public frames

Categories : pixel-level semantic annotations for 20 classes

Code/Model/Publication URL : WildDash – Creating Hazard-Aware Benchmarks

Source URL : WildDash 2 Dataset

Dataset Examples :

Credits : WildDash – Creating Hazard-Aware Benchmarks

 

72. CAD 120 affordance dataset

      Dataset Characteristics :

Task Type semantic image segmentation

Image Attribute contain 3090 images containing 9916 object instances

The images contain the following object classes:
1.table
2.kettle
3.plate
4.bottle
5.thermal cup
6.knife
7.medicine box
8.can
9.microwave
10.paper box
11.bowl
12.mug

Affordances in the dataset:
1.openable
2.cuttable
3.pourable
4.containable
5.supportable
6.holdable

Code/Model/Publication URL: Weakly Supervised Affordance Detection

Source URL : CAD 120 Affordance Segmentation Dataset

Dataset Examples :

Credits : Weakly Supervised Affordance Detection

 

73. The Aberystwyth Leaf Evaluation Dataset

      Dataset Characteristics :

Task Type Leaf Segmentation

Image Attribute4 sets of 20 Arabidopsis Thaliana plants have been grown in trays Images of each tray are taken in a 15 minute timelapse sequence using a robotic greenhouse system.

There are 56 annotated ground truth images containing 916 hand-marked up individual arabidopsis plants.

Code/Model/Publication URL : The Aberystwyth Leaf Evaluation Dataset: A plant growth visible light image dataset of Arabidopsis thalian

Source URL : The Aberystwyth Leaf Evaluation Dataset

Dataset Examples :

Credits : The Aberystwyth Leaf Evaluation Dataset: A plant growth visible light image dataset of Arabidopsis thalian

 

74. Shadow Detection/Texture Segmentation Computer Vision Dataset

      Dataset Characteristics :

Task Type Texture Segmentation

Image Attribute : 53 outdoor images with objects influenced by shadows

Code/Model/Publication URL : Shadow free segmentation in still images using local density measure

Source URL : Shadow Detection/Texture Segmentation Computer Vision Dataset

Dataset Examples :

Credits : Shadow free segmentation in still images using local density measure

 

75. CITY-OSM – ETH Zurich Dataset

      Dataset Characteristics :

Task Type Aerial Image Segmentation

Image AttributeFour large datasets were downloaded from Google Maps and OSM, for the cities of Chicago, Paris, Zurich, and Berlin.Additionally and also downloaded a somewhat smaller dataset for the city of Potsdam.

Code/Model/Publication URL : Learning Aerial Image Segmentation from Online Maps

Source URL : CITY-OSM – ETH Zurich Dataset

Dataset Examples :

Credits : Learning Aerial Image Segmentation from Online Maps

 

76.  The TUD Crossing dataset

       Dataset Characteristics :

Task Type pedestrian instance segmentation

Image Attribute : 201 images  with 1008 highly overlapping pedestrians and consisting of 1216 pedestrian instances with densely segmented overlapping pedestrians.

Code/Model/Publication URL : Hough Regions for Joining Instance Localization and Segmentation

Source URL : The TUD Crossing dataset

Dataset Examples :

Credits : Hough Regions for Joining Instance Localization and Segmentation

 

77. Inria Aerial Image Labeling Dataset

       Dataset Characteristics :

Task Type semantic segmentation on aerial images

Image Attribute : Coverage of 810 km² (405 km² for training and 405 km² for testing) having Ground truth data for two semantic classes: building and not building

Code/Model/Publication URL : CAN SEMANTIC LABELING METHODS GENERALIZE TO ANY CITY?THE INRIA AERIAL IMAGE LABELING BENCHMARK

Source URL : Inria Aerial Image Labeling Dataset

Dataset Examples :

Credits : CAN SEMANTIC LABELING METHODS GENERALIZE TO ANY CITY?THE INRIA AERIAL IMAGE LABELING BENCHMARK

 

78.  Zurich Summer Dataset

       Dataset Characteristics :

Task Type : Semantic segmentation of urban scene

Image Attribute : 20 multi-spectral VHR images acquired over the city of Zurich(Switzerland) by the QuickBird satellite in 2002. The average image size is 1000×1150 pixels (approximately 23Mpixels in total)

Categories : 8 different urban and periurban classes : Roads, Buildings, Trees, Grass, Bare Soil, Water, Railways and Swimming pools

Code/Model/Publication URL : Semantic segmentation of urban scenes by learning local class interactions

Source URL : Zurich Summer Dataset

Dataset Examples :

Credits : Semantic segmentation of urban scenes by learning local class interactions

 

79. Pedestrian Color Naming Dataset

      Dataset Characteristics :

Task Type : image segmentation

Image Attribute14,213 images, each of which hand-labeled with color label for each pixel

Code/Model/Publication URL : Pedestrian Color Naming via Convolutional Neural Network

Source URL : Pedestrian Color Naming Dataset

Dataset Examples :

Credits : Pedestrian Color Naming via Convolutional Neural Network

 

80. Human Skin Segmentation Dataset

      Dataset Characteristics :

Task Type : human skin segmentation

Image Attribute : 32 face photo, 46 family photo with ground truth labels

Code/Model/Publication URL : A Fusion Approach for Efficient Human Skin Detection

Source URL : Human Skin Segmentation Dataset

Dataset Examples :

Credits : A Fusion Approach for Efficient Human Skin Detection

 

81. 38-Cloud: A Cloud Segmentation Dataset

      Dataset Characteristics :

Task Type Cloud Segmentation

Image Attribute : 8400 patches for training and 9201 patches for testing extracted from 38 Landsat 8 remote sensing images

Code/Model/Publication URL : CLOUD-NET: AN END-TO-END CLOUD DETECTION ALGORITHM FOR LANDSAT 8 IMAGERY

Source URL : 38-Cloud-A-Cloud-Segmentation-Dataset

Dataset Examples :

Credits : 38-Cloud-A-Cloud-Segmentation-Dataset

 

82. Multi-Human Parsing (MHP) v2.0 Dataset

      Dataset Characteristics :

Task Type : multi-human parsing

Image Attribute25,403 human images with pixel-wise annotations

Categories : 58 semantic categories

cap/hathelmet
facehair
left-armright-arm
left-handright-hand
protectorbikini/bra
jacket/windbreaker/hoodiet-shirt
polo-shirtsweater
singlettorso-skin
pantsshorts/swim-shorts
skirtstockings
socksleft-boot
right-bootleft-shoe
right-shoeleft-high heel
right-high heelleft-sandal
right-sandalleft-leg
right-legleft-foot
right-footcoat
dressrobe
jumpsuitother-full-body-clothes
headwearbackpack
ballbats
beltbottle
carrybagcases
sunglasseseyewear
glovescarf
umbrellawallet/purse
watchwristband
tieother-accessory
other-upper-body-clothesother-lower-body-clothes

Code/Model/Publication URL : Understanding Humans in Crowded Scenes: Deep Nested Adversarial Learning and A New Benchmark for Multi-Human Parsing

Source URL : Multi-Human Parsing (MHP) v2.0 Dataset

Dataset Examples :

Credits : Understanding Humans in Crowded Scenes: Deep Nested Adversarial Learning and A New Benchmark for Multi-Human Parsing

 

83. MSRA10K Salient Object Database

      Dataset Characteristics :

Task Type : salient object detection and segmentation

Image Attribute10,000 images with pixel-level saliency labeling + Pixel accurate salient object labeling for 5000 images from MSRA-B dataset

Code/Model/Publication URL : Global Contrast based Salient Region Detection

Source URL : MSRA10K Salient Object Database

Dataset Examples :

Credits : MSRA10K Salient Object Database

 

84. Stanford background dataset

      Dataset Characteristics :

Task Type : geometric and semantic scene understanding.

Image Attribute : 715 images have approximately 320-by-240 pixels.

Categories :

  • sky
  • tree
  • road
  • grass
  • water
  • bldg
  • mntn
  • fg obj.
  • horz.
  • vert.

Code/Model/Publication URL : Decomposing a Scene into Geometric and Semantically Consistent Regions

Source URL : Stanford background dataset

Dataset Examples :

Credits : Decomposing a Scene into Geometric and Semantically Consistent Regions

 

85. Oakland 3-D Point Cloud Dataset

      Dataset Characteristics :

Task Type : semantic segmentation

Image Attribute17 files, 1.6 millions 3-D pts, 44 labels

Code/Model/Publication URL : Onboard Contextual Classification of 3-D Point Clouds with Learned High-order Markov Random Fields

Source URL : Oakland 3-D Point Cloud Dataset

Dataset Examples :

Credits : Onboard Contextual Classification of 3-D Point Clouds with Learned High-order Markov Random Fields

 

86. Penn-Fudan Database for Pedestrian Detection and Segmentation

      Dataset Characteristics :

Task Type : pedestrian segmentation

Image Attribute170 images with 345 labeled pedestrians, among which 96 images are taken from around University of Pennsylvania, and other 74 are taken from around Fudan University

Code/Model/Publication URL : Object Detection Combining Recognition and Segmentation

Source URL : Penn-Fudan Database for Pedestrian Detection and Segmentation

Dataset Examples :

Credits : Penn-Fudan Database for Pedestrian Detection and Segmentation

 

87. HOPKINS 155 DATASET

      Dataset Characteristics :

Task Type : motion segmentation

Image Attribute : 155 motion sequences of checkerboard, traffic, and articulated scenes

Code/Model/Publication URL : A Benchmark for the Comparison of 3-D Motion Segmentation Algorithms

Source URL : HOPKINS 155 DATASET

Dataset Examples :

Credits : HOPKINS 155 DATASET

 

88. Segmentation evaluation database

      Dataset Characteristics :

Task Type : image segmentation

Image Attribute : 200 gray level images along with ground truth segmentations

Code/Model/Publication URL : Image Segmentation by Probabilistic Bottom-Up Aggregation and Cue Integration

Source URL : Segmentation evaluation database

Dataset Examples :

Credits : Image Segmentation by Probabilistic Bottom-Up Aggregation and Cue Integration

 

89. CO-SKEL dataset

      Dataset Characteristics :

Task Type : co-skeletonization, Co-segmentation

Image Attribute : It consists of 26 categories with total 353 images of animals,birds, flowers and humans

beariris
camelcat
cheetahcormorant
cowcranesbill
deerdesertrose
dogegret
firepinkfrog
geraniumhorse
manostrich
pandapigeon
seagullseastar
sheepsnowowl
statuewoman

Code/Model/Publication URL : Object Co-skeletonization with Co-segmentation

Source URL : CO-SKEL dataset

Dataset Examples :

Credits : Object Co-skeletonization with Co-segmentation

 

90. EV-IMO dataset

      Dataset Characteristics :

Task Type : indoor motion segmentation

Video Attribute

  • Total recording time ~30 minutes
  • Up to 3 independently moving objects objects
  • Multiple types of scenes (varying backgrounds and motion speeds)

Code/Model/Publication URL : EV-IMO: Motion Segmentation Dataset and Learning Pipeline for Event Cameras

Source URL : EVIMO dataset

Dataset Examples :

Credits : EVIMO dataset

 

91. Materials in Context Database (MINC) dataset

      Dataset Characteristics :

Task Type : material segmentation of images in the wild.

Image AttributeIt consists of 3M labeled point samples and 7061 labeled material segmentations in 23 material categories.

BrickCarpet
CeramicFabric
FoliageFood
GlassHair
LeatherMirror
MetalOther
PaintedPaper
PlasticSkin
Pol. stoneSky
StoneTile
WallpaperWood
Water

Code/Model/Publication URL : Material Recognition in the Wild with the Materials in Context Database

Source URL : MINC dataset

Dataset Examples :

Credits : Material Recognition in the Wild with the Materials in Context Database

 

92. Liver Tumor Segmentation (LiTS) dataset

      Dataset Characteristics :

Task Type : Liver Tumor Segmentation

Image Attribute : the training data set contains 130 CT scans and the test data set 70 CT scans

Code/Model/Publication URL : The Liver Tumor Segmentation Benchmark (LiTS)

Source URL : (LiTS) dataset

Dataset Examples :

Credits : The Liver Tumor Segmentation Benchmark (LiTS)

 

93.  Egocentric Dataset of the University of Barcelona – Segmentation (EDUB-Seg) dataset

      Dataset Characteristics :

Task Type egocentric event segmentation

Image Attributea total of 18,735 images captured by 7 different users during overall 20 days.

Code/Model/Publication URL : R-Clustering for Egocentric Video Segmentation

Source URL : EDUB-Seg dataset

Dataset Examples :

Credits : R-Clustering for Egocentric Video Segmentation

 

94. Audio Visual Cues Dataset

      Dataset Characteristics :

Task Type : semantic segmentation from audio-visual clues

Image Attribute9 long reconstruction sequences, containing on average 1600 individual 640×480 RGB-D frames. 600 sounds from 50 different objects and 9 material categories were collected.

Code/Model/Publication URL : Joint Object-Material Category Segmentation from Audio-Visual Cues

Source URL : Audio Visual Cues Dataset

Dataset Examples :

Credits : Joint Object-Material Category Segmentation from Audio-Visual Cues

 

95. OpenSurfaces Dataset

      Dataset Characteristics :

Task Type : surfaces segmentation from consumer photographs of indoor scene

Image Attribute25K images filtered then 58,928 surfaces annotated with a material name, and 33,378 annotated with object names.

Code/Model/Publication URL : OPENSURFACES: A Richly Annotated Catalog of Surface Appearance

Source URL : OpenSurfaces Dataset

Dataset Examples :

Credits : OPENSURFACES: A Richly Annotated Catalog of Surface Appearance

 

96.  Multi-species fruit flower detection

       Dataset Characteristics :

Task Type : semantic segmentation

Image Attribute : 190 images shared among three different species: apple, peach, and pear

Code/Model/Publication URL : Multispecies Fruit Flower Detection Using a Refined Semantic Segmentation Network

Source URL : Data from: Multi-species fruit flower detection using a refined semantic segmentation network

Dataset Examples :

Credits : Multispecies Fruit Flower Detection Using a Refined Semantic Segmentation Network

 

97. SAIL-VOS

      Dataset Characteristics :

Task Type : video object segmentation

Image Attribute : 201 video sequences and 111,654 frames consisting of 1,388,389 objects

Code/Model/Publication URL : SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation –A Synthetic Dataset and Baselines

Source URL : SAIL-VOS

Dataset Examples :

Credits : SAIL-VOS: Semantic Amodal Instance Level Video Object Segmentation –A Synthetic Dataset and Baselines

 

98. TB-roses-v1 & v2 dataset

      Dataset Characteristics :

Task Type : rose stem segmentation

Image Attribute : 354 images of rose bushes

Code/Model/Publication URL : Brain-inspired robust delineation operator

Source URL : TB-roses-v1 & v2 dataset

Dataset Examples :

Credits : TB-roses-v1 & v2 dataset

99. IOSTAR Retinal Vessel Segmentation Dataset

      Dataset Characteristics :

Task Type Vessel Segmentation

Image Attribute : 30 images with a resolution of 1024×1024 pixels

Code/Model/Publication URL : Robust Retinal Vessel Segmentation via Locally Adaptive Derivative Frames in Orientation Scores

Source URL : IOSTAR Retinal Vessel Segmentation Dataset

*require institutional login

Dataset Examples :

Credits : Robust Retinal Vessel Segmentation via Locally Adaptive Derivative Frames in Orientation Scores

 

100.  PartNet Dataset

        Dataset Characteristics :

Task Type : fine-grained semantic segmentation, hierarchical semantic segmentation, and instance segmentation

Image Attribute : 573,585 part instances over 26,671 3D models covering 24 object categories.

24 object categories in PartNet

Code/Model/Publication URL : PartNet: A Large-scale Benchmark for Fine-grained and Hierarchical Part-level 3D Object Understanding 

Source URL : PartNet Dataset

Dataset Examples :

Credits : PartNet Dataset

 

101. Synthinel-1 dataset

        Dataset Characteristics :

Task Type : Building footprint segmentation

Image Attribute : 2,108 synthetic images with Each synthetic image is 572×572 pixels in size, with a resolution 0.3m/pixel

The nine different virtual city styles used are.

(a) Red roof style

(b) Paris’ buildings style

(c) ancient building style

(d) sci-fi city style

(e) Chinese palace style

(f) Damaged city style

(g) Austin city style

(h) Venice style

(i) modern city style

Code/Model/Publication URL : The Synthinel-1 dataset: a collection of high resolution synthetic overhead imagery for building segmentation 

Source URL : Synthinel-1 dataset

Dataset Examples :

Credits : Synthinel-1 dataset


Do you have a custom dataset with several object classes for which you desire pixel-perfect annotations ?

Well you have reached the right place, hop on to NeuralMarker An-End-to-End AI Annotation Platform in order to Label, Manage and Utilize it’s capabilities for creating your next pixel-perfect quality training data.

Want to know which AI Tools NeuralMarker An-End-to-End AI Annotation Platform  uses to segment pixels over the images. Click here to know more.

Leave a Comment

Your email address will not be published. Required fields are marked *