Self-Supervised Video Representation Learning With Odd-One-Out Networks

Abstract

We propose a new self-supervised CNN pre-training technique based on a novel auxiliary task called "odd-one-out learning". In this task, the machine is asked to identify the unrelated or odd element from a set of otherwise related elements. We apply this technique to self-supervised video representation learning where we sample subsequences from videos and ask the network to learn to predict the odd video subsequence. The odd video subsequence is sampled such that it has wrong temporal order of frames while the even ones have the correct temporal order. Therefore, to generate a odd-one-out question no manual annotation is required. Our learning machine is implemented as multi-stream convolutional neural network, which is learned end-to-end. Using odd-one-out networks, we learn temporal representations for videos that generalizes to other related tasks such as action recognition. On action classification, our method obtains 60.3\% on the UCF101 dataset using only UCF101 data for training which is approximately 10% better than current state-of-the-art self-supervised learning methods. Similarly, on HMDB51 dataset we outperform self-supervised state-of-the art methods by 12.7% on action classification task.

Accepted in In IEEE International Conference on Computer Vision and Pattern Recognition CVPR 2017

References in corpus (13)

• Sequence to Sequence Learning with Neural Networks
• UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
• YouTube-8M: A Large-Scale Video Classification Benchmark
• MatConvNet - Convolutional Neural Networks for MATLAB
• Rank Pooling for Action Recognition
• Unsupervised Visual Representation Learning by Context Prediction
• Unsupervised Learning of Visual Representations using Videos
• Learning Local Image Descriptors with Deep Siamese and Triplet Convolutional Networks by Minimising Global Loss Functions
• Learning Spatiotemporal Features with 3D Convolutional Networks
• Learning to See by Moving
• Human Action Recognition using Factorized Spatio-Temporal Convolutional Networks
• VideoLSTM Convolves, Attends and Flows for Action Recognition
• Action Recognition with Dynamic Image Networks

Cited by in corpus (5)

• Self-supervised Co-training for Video Representation Learning
• Memory-augmented Dense Predictive Coding for Video Representation Learning
• DiscrimNet: Semi-Supervised Action Recognition from Videos using Generative Adversarial Networks
• A Large-Scale Study on Unsupervised Spatiotemporal Representation Learning
• Generating Music Medleys via Playing Music Puzzle Games