Cycle-Contrast for Self-Supervised Video Representation Learning

Abstract

We present Cycle-Contrastive Learning (CCL), a novel self-supervised method for learning video representation. Following a nature that there is a belong and inclusion relation of video and its frames, CCL is designed to find correspondences across frames and videos considering the contrastive representation in their domains respectively. It is different from recent approaches that merely learn correspondences across frames or clips. In our method, the frame and video representations are learned from a single network based on an R3D architecture, with a shared non-linear transformation for embedding both frame and video features before the cycle-contrastive loss. We demonstrate that the video representation learned by CCL can be transferred well to downstream tasks of video understanding, outperforming previous methods in nearest neighbour retrieval and action recognition tasks on UCF101, HMDB51 and MMAct.

12 pages

References in corpus (11)

• UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
• The Kinetics Human Action Video Dataset
• Unsupervised Representation Learning by Predicting Image Rotations
• A Structured Self-attentive Sentence Embedding
• Generating Videos with Scene Dynamics
• Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning
• Learning Video Representations using Contrastive Bidirectional Transformer
• Audiovisual SlowFast Networks for Video Recognition
• Contrastive Adaptation Network for Unsupervised Domain Adaptation
• Deep generative-contrastive networks for facial expression recognition
• Shuffle and Learn: Unsupervised Learning using Temporal Order Verification

Cited by in corpus (5)

• Dual Contrastive Learning for Spatio-temporal Representation
• Contrastive Learning of Person-independent Representations for Facial Action Unit Detection
• Motion-aware Contrastive Video Representation Learning via Foreground-background Merging
• Contrastive Learning of Global-Local Video Representations
• Contrastive Learning of Image Representations with Cross-Video Cycle-Consistency