Caltech Pedestrian

The dataset used in the paper is a video prediction dataset with occlusions, which is used to evaluate the proposed Fast Fourier Inception Networks (FFINet) for occluded video prediction.

BibTex: