DAVIS Object Segmentation Dataset

Densely Annotation Video Segmentation dataset (DAVIS) is a high-quality, high-resolution video object segmentation dataset with two resolutions, 480p and 1080p. The dataset includes 50 videos and 3,455 densely annotated frames at the pixel level. The training set contains 30 videos (2079 frames); the validation set contains 20 videos (1376 frames).

Data example: