Rethinking the Faster R-CNN Architecture for Temporal Action Localization

• Published in: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition pp. 1130 - 1139
• Main Authors: Chao, Yu-Wei, Vijayanarasimhan, Sudheendra, Seybold, Bryan, Ross, David A., Deng, Jia, Sukthankar, Rahul
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2018
• Subjects: Computer architecture; Feature extraction; Image segmentation; Microsoft Windows; Object detection; Proposals; Two dimensional displays

We propose TAL-Net, an improved approach to temporal action localization in video that is inspired by the Faster RCNN object detection framework. TAL-Net addresses three key shortcomings of existing approaches: (1) we improve receptive field alignment using a multi-scale architecture that can accomm...