3C-Net: Category Count and Center Loss for Weakly-Supervised Action Localization

TL;DR

3C-Net is a weakly-supervised action localization framework that uses video-level labels and counts to learn discriminative features, achieving state-of-the-art results on THUMOS14 and ActivityNet 1.2 datasets.

Contribution

It introduces a novel joint formulation combining classification, multi-label center loss, and counting loss for improved localization with weak supervision.

Findings

Achieves 4.6% mAP improvement on THUMOS14.

Sets new state-of-the-art on both THUMOS14 and ActivityNet 1.2.

Effectively delineates adjacent action sequences with counting loss.

Abstract

Temporal action localization is a challenging computer vision problem with numerous real-world applications. Most existing methods require laborious frame-level supervision to train action localization models. In this work, we propose a framework, called 3C-Net, which only requires video-level supervision (weak supervision) in the form of action category labels and the corresponding count. We introduce a novel formulation to learn discriminative action features with enhanced localization capabilities. Our joint formulation has three terms: a classification term to ensure the separability of learned action features, an adapted multi-label center loss term to enhance the action feature discriminability and a counting loss term to delineate adjacent action sequences, leading to improved localization. Comprehensive experiments are performed on two challenging benchmarks: THUMOS14 and ActivityNet 1.2. Our approach sets a new state-of-the-art for weakly-supervised temporal action localization on both datasets. On the THUMOS14 dataset, the proposed method achieves an absolute gain of 4.6% in terms of mean average precision (mAP), compared to the state-of-the-art. Source code is available at https://github.com/naraysa/3c-net.