Weakly-guided Self-supervised Pretraining for Temporal Activity Detection

TL;DR

This paper introduces a weakly-guided self-supervised pretraining approach for temporal activity detection that leverages classification data to generate pseudo labels, improving detection performance without requiring additional annotations.

Contribution

The authors propose a novel self-supervised pretraining method that uses weak labels to create a detection-oriented pretext task, bridging the gap between classification pretraining and detection fine-tuning.

Findings

Outperforms previous methods on Charades and MultiTHUMOS benchmarks.

Effectively generates frame-level pseudo labels from classification data.

Provides insights on optimal usage of the proposed models for activity detection.

Abstract

Temporal Activity Detection aims to predict activity classes per frame, in contrast to video-level predictions in Activity Classification (i.e., Activity Recognition). Due to the expensive frame-level annotations required for detection, the scale of detection datasets is limited. Thus, commonly, previous work on temporal activity detection resorts to fine-tuning a classification model pretrained on large-scale classification datasets (e.g., Kinetics-400). However, such pretrained models are not ideal for downstream detection, due to the disparity between the pretraining and the downstream fine-tuning tasks. In this work, we propose a novel 'weakly-guided self-supervised' pretraining method for detection. We leverage weak labels (classification) to introduce a self-supervised pretext task (detection) by generating frame-level pseudo labels, multi-action frames, and action segments. Simply put, we design a detection task similar to downstream, on large-scale classification data, without extra annotations. We show that the models pretrained with the proposed weakly-guided self-supervised detection task outperform prior work on multiple challenging activity detection benchmarks, including Charades and MultiTHUMOS. Our extensive ablations further provide insights on when and how to use the proposed models for activity detection. Code is available at https://github.com/kkahatapitiya/SSDet.