TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Weakly Supervised Action Localization	THUMOS14	TSCN	avg-mAP (0.1-0.5)	47.0	# 10
Weakly Supervised Action Localization	THUMOS14	TSCN	avg-mAP (0.3-0.7)	28.8	# 12
Weakly Supervised Action Localization	THUMOS14	TSCN	avg-mAP (0.1:0.7)	37.8	# 10

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/two-stream-consensus-network-for-weakly-1/weakly-supervised-action-localization-on-5)](https://paperswithcode.com/sota/weakly-supervised-action-localization-on-5?p=two-stream-consensus-network-for-weakly-1)`

Two-Stream Consensus Network for Weakly-Supervised Temporal Action Localization

ECCV 2020 · Yuanhao Zhai, Le Wang, Wei Tang, Qilin Zhang, Junsong Yuan, Gang Hua ·

Weakly-supervised Temporal Action Localization (W-TAL) aims to classify and localize all action instances in an untrimmed video under only video-level supervision. However, without frame-level annotations, it is challenging for W-TAL methods to identify false positive action proposals and generate action proposals with precise temporal boundaries. In this paper, we present a Two-Stream Consensus Network (TSCN) to simultaneously address these challenges. The proposed TSCN features an iterative refinement training method, where a frame-level pseudo ground truth is iteratively updated, and used to provide frame-level supervision for improved model training and false positive action proposal elimination. Furthermore, we propose a new attention normalization loss to encourage the predicted attention to act like a binary selection, and promote the precise localization of action instance boundaries. Experiments conducted on the THUMOS14 and ActivityNet datasets show that the proposed TSCN outperforms current state-of-the-art methods, and even achieves comparable results with some recent fully-supervised methods.