Center for Research in Comptuer Vision
Center for Research in Comptuer Vision



An End-to-end 3D Convolutional Neural Network for Action Detection and Segmentation in Videos


Results



Qualitative results of the proposed approach (red), ARP (yellow), LVO (cyan) and FSEG (magenta) on selected frames from DAVIS dataset.







Example segmentation results of the proposed approach for a few testing video sequences from DAVIS dataset.









Code

Coming soon.

Related Publication

R. Hou, C. Chen, and M. Shah, An End-to-end 3D Convolutional Neural Network for Action Detection and Segmentation in Videos, arXiv:1712.01111 [cs.CV], 2017.

Back to Human Action and Activity Recognition Projects