Cross-Architecture Self-supervised Video Representation Learning

TL;DR

This paper introduces a cross-architecture contrastive learning framework combining 3D CNNs and video transformers, along with a temporal self-supervised module, to improve self-supervised video representation learning, achieving state-of-the-art results on standard datasets.

Contribution

The paper proposes a novel cross-architecture contrastive learning framework and a temporal self-supervised module for enhanced video representation learning.

Findings

Outperforms state-of-the-art methods on UCF101 and HMDB51 datasets.

Achieves significant improvements in video retrieval and action recognition tasks.

Demonstrates the effectiveness of combining 3D CNNs and transformers in self-supervised learning.

Abstract

In this paper, we present a new cross-architecture contrastive learning (CACL) framework for self-supervised video representation learning. CACL consists of a 3D CNN and a video transformer which are used in parallel to generate diverse positive pairs for contrastive learning. This allows the model to learn strong representations from such diverse yet meaningful pairs. Furthermore, we introduce a temporal self-supervised learning module able to predict an Edit distance explicitly between two video sequences in the temporal order. This enables the model to learn a rich temporal representation that compensates strongly to the video-level representation learned by the CACL. We evaluate our method on the tasks of video retrieval and action recognition on UCF101 and HMDB51 datasets, where our method achieves excellent performance, surpassing the state-of-the-art methods such as VideoMoCo and MoCo+BE by a large margin. The code is made available at https://github.com/guoshengcv/CACL.