Adversarial Background-Aware Loss for Weakly-supervised Temporal Activity Localization

ECCV 2020  ·  Kyle Min, Jason J. Corso ·

Temporally localizing activities within untrimmed videos has been extensively studied in recent years. Despite recent advances, existing methods for weakly-supervised temporal activity localization struggle to recognize when an activity is not occurring. To address this issue, we propose a novel method named A2CL-PT. Two triplets of the feature space are considered in our approach: one triplet is used to learn discriminative features for each activity class, and the other one is used to distinguish the features where no activity occurs (i.e. background features) from activity-related features for each video. To further improve the performance, we build our network using two parallel branches which operate in an adversarial way: the first branch localizes the most salient activities of a video and the second one finds other supplementary activities from non-localized parts of the video. Extensive experiments performed on THUMOS14 and ActivityNet datasets demonstrate that our proposed method is effective. Specifically, the average mAP of IoU thresholds from 0.1 to 0.9 on the THUMOS14 dataset is significantly improved from 27.9% to 30.0%.

PDF Abstract ECCV 2020 PDF ECCV 2020 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Weakly Supervised Action Localization ActivityNet-1.3 A2CL-PT mAP@0.5 36.8 # 5
mAP@0.5:0.95 22.5 # 5
Weakly Supervised Action Localization THUMOS14 A2CL-PT avg-mAP (0.1-0.5) 46.9 # 9
avg-mAP (0.3-0.7) 30.6 # 8
avg-mAP (0.1:0.7) 37.8 # 8
Weakly-supervised Temporal Action Localization THUMOS’14 A2CL-PT mAP IOU@0.1 61.2 # 2
mAP IOU@0.2 56.1 # 2
mAP IOU@0.3 48.1 # 2
mAP IOU@0.4 39.0 # 2
mAP IOU@0.5 30.1 # 2
mAP IOU@0.6 19.2 # 2
mAP IOU@0.7 10.6 # 2
mAP IOU@0.8 4.8 # 2
mAP IOU@0.9 1.0 # 2
mAP@AVG(0.1:0.9) 30.0 # 2
Weakly Supervised Action Localization THUMOS’14 A2CL-PT mAP@0.5 30.1 # 9
Weakly Supervised Action Localization THUMOS 2014 A2CL-PT mAP@0.5 30.1 # 10
mAP@0.1:0.7 37.8 # 9
mAP@0.1:0.5 46.9 # 9

Methods


No methods listed for this paper. Add relevant methods here