Look into Person (LIP) is a new large-scale dataset, focus on semantic understanding of person. Following are the detailed descriptions.
The dataset contains 50,000 images with elaborated pixel-wise annotations with 19 semantic human part labels and 2D human poses with 16 key points.
The annotated 50,000 images are cropped person instances from COCO dataset with size larger than 50 * 50.The images collected from the real-world scenarios contain human appearing with challenging poses and views, heavily occlusions, various appearances and low-resolutions. We are working on collecting and annotating more images to increase diversity.
2.1 Single Person
Besides we have another large dataset mentioned in "Human parsing with contextualized convolutional neural network." ICCV'15, which focuses on fashion images. You can download the dataset including 17000 images as extra training data.
To stimulate the multiple-human parsing research, we collect the images with multiple person instances in the LIP dataset to establish the first standard and comprehensive benchmark for multiple-human parsing and pose estimation. Our LIP Multiple-Human Parsing Dataset contains 28280 training, 5000 validation and 5000 test images, in which there are 38280 multiple-person images in total.
2.3 Video Multi-Person Human Parsing
VIP(Video instance-level Parsing) dataset, the first video multi-person human parsing benchmark, consists of 404 videos covering various scenarios. For every 25 consecutive frames in each video, one frame is annotated densely with pixel-wise semantic part categories and instance-level identification. There are 21247 densely annotated images in total. We divide these 404 sequences into 304 train sequences, 50 validation sequences and 50 test sequences.
- VIP_Fine: All annotated images and fine annotations for train and val sets.
- VIP_Sequence: 20-frame surrounding each VIP_Fine image (-10 | +10).
- VIP_Videos: 404 video sequences of VIP dataset.