SeCo: Exploring Sequence Supervision for Unsupervised Representation Learning
This work addresses the problem of unsupervised learning for video analysis, offering a novel method that improves performance on tasks like action recognition and object tracking, though it is incremental in building on contrastive learning paradigms.
The paper tackles unsupervised representation learning for video by leveraging inherent sequential supervision, proposing Sequence Contrastive Learning (SeCo) which achieves superior results, including leading accuracy by 2.96% and 6.47% over fully-supervised ImageNet pre-training on UCF101 and HMDB51 action recognition tasks.
A steady momentum of innovations and breakthroughs has convincingly pushed the limits of unsupervised image representation learning. Compared to static 2D images, video has one more dimension (time). The inherent supervision existing in such sequential structure offers a fertile ground for building unsupervised learning models. In this paper, we compose a trilogy of exploring the basic and generic supervision in the sequence from spatial, spatiotemporal and sequential perspectives. We materialize the supervisory signals through determining whether a pair of samples is from one frame or from one video, and whether a triplet of samples is in the correct temporal order. We uniquely regard the signals as the foundation in contrastive learning and derive a particular form named Sequence Contrastive Learning (SeCo). SeCo shows superior results under the linear protocol on action recognition (Kinetics), untrimmed activity recognition (ActivityNet) and object tracking (OTB-100). More remarkably, SeCo demonstrates considerable improvements over recent unsupervised pre-training techniques, and leads the accuracy by 2.96% and 6.47% against fully-supervised ImageNet pre-training in action recognition task on UCF101 and HMDB51, respectively. Source code is available at \url{https://github.com/YihengZhang-CV/SeCo-Sequence-Contrastive-Learning}.