The code is based on C3D v-1.1.
Models are downloaded here.
Pre-computed results are downloaded from our project page.
Please cite the following paper:
`@article{ltnghia-2017,
author = {Trung-Nghia Le and Akihiro Sugimoto,
title = {Deeply Supervised 3D Recurrent FCN for Salient Object Detection in Videos},
journal = {BMVC},
year = {2017}
}`
Contact: Trung-Nghia Le.