Scale Matters: Temporal Scale Aggregation Network for Precise Action Localization in Untrimmed Videos
arXiv:1908.00707
The paper introduces TSA-Net, a network that uses multi‑dilation temporal convolutions and multiple branches to better capture actions of varying lengths and to detect precise start, middle, and end points for action proposals in untrimmed videos.
Abstract
Temporal action localization is a recently-emerging task, aiming to localize video segments from untrimmed videos that contain specific actions. Despite the remarkable recent progress, most two-stage action localization methods still suffer from imprecise temporal boundaries of action proposals. This work proposes a novel integrated temporal scale aggregation network (TSA-Net). Our main insight is that ensembling convolution filters with different dilation rates can effectively enlarge the receptive field with low computational cost, which inspires us to devise multi-dilation temporal convolution (MDC) block. Furthermore, to tackle video action instances with different durations, TSA-Net consists of multiple branches of sub-networks. Each of them adopts stacked MDC blocks with different dilation parameters, accomplishing a temporal receptive field specially optimized for specific-duration actions. We follow the formulation of boundary point detection, novelly detecting three kinds of critical points (ie, starting / mid-point / ending) and pairing them for proposal generation. Comprehensive evaluations are conducted on two challenging video benchmarks, THUMOS14 and ActivityNet-1.3. Our proposed TSA-Net demonstrates clear and consistent better performances and re-calibrates new state-of-the-art on both benchmarks. For example, our new record on THUMOS14 is 46.9% while the previous best is 42.8% under mAP@0.5.