Datasets¶
wiki: List of datasets for machine learning research
Image¶
ImageNet¶
14 million images
20000 categories
Labeled objects, bounding boxes, descriptive words, SIFT features
CIFAR-10/100¶
CIFAR-10 and CIFAR-100 datasets
low-resolution. Good for generative model. Difficult to learn comparing to face
CIFAR-10: 10 classes, with 6000 images per class
CIFAR-100: 100 classes containing 600 images each
NUS-WIDE¶
Common Objects in COntext (COCO)¶
COCO - Complex Adaptive Systems Laboratory
Number of images in the dataset: 330,000 images while more than 200,000 are labeled (roughly equal halves for training and validation+test)
Number of classes: 80 object categories, 91 stuff categories
Image resolution: 640×480
Open Image¶
Open Image Dataset V5 by google label + boxes + segmentation + relationship annotation
Image Segmenation¶
Images | Obj. Inst | Obj. Cls | Part Inst. | Part Cls | Obj. Cls. per Img | |
---|---|---|---|---|---|---|
COCO | 123,287 | 886,284 | 91 | 0 | 0 | 3.5 |
ImageNet∗ | 476,688 | 534,309 | 200 | 0 | 0 | 1.7 |
NYU Depth V2 | 1,449 | 34,064 | 894 | 0 | 0 | 14.1 |
Cityscapes | 25,000 | 65,385 | 30 | 0 | 0 | 12.2 |
SUN | 16,873 | 313,884 | 4,479 | 0 | 0 | 9.8 |
OpenSurfaces | 22,214 | 71,460 | 160 | 0 | 0 | N/A |
PascalContext | 10,103 | ∼104,398∗∗ | 540 | 181,770 | 40 | 5.1 |
ADE20K | 22,210 | 434,826 | 2,693 | 175,961 | 476 | 9.9 |
from Scene Parsing through ADE20K Dataset |
RGB-D¶
SUN RGB-D¶
A RGB-D Scene Understanding Benchmark Suite
Project page
Challenge
NYU dataset¶
https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html
Object Tracking¶
youtube-bb¶
MOT Challenge: Multiple Object Tracking¶
MOT Challenge detection is provided
Pose¶
Human Activity¶
Video¶
Optical Flow¶
Video Debluring¶
- Hand-held camaera from Deep Video Deblurring for Hand-held Cameras (CVPR 2017)
- GoPro, from Deep Multi-Scale Convolutional Neural Network for Dynamic Scene Deblurring (CVPR 2017)
- REDS from NTIRE 2019
Video Restoration¶
REDS¶
from NTIRE 2019
- sharp (ground truth)
- blur
- blur+compression
- low resolution
- blur + low resolution
Vimeo90K¶
from Video Enhancement with Task-Oriented Flow (IJCV 2019)
- temporal frame interpolation
- video denoising
- video deblocking
- video super-resolution
Image grading¶
MIT-Adobe FiveK Dataset¶
Adobe FiveK 5,000 photos in DNG format An Adobe Lightroom catalog with renditions by 5 experts Semantic information about each photo