Revisiting Contrastive Methods for Unsupervised Learning of Visual Representations

Abstract

Contrastive self-supervised learning has outperformed supervised pretraining on many downstream tasks like segmentation and object detection. However, current methods are still primarily applied to curated datasets like ImageNet. In this paper, we first study how biases in the dataset affect existing methods. Our results show that current contrastive approaches work surprisingly well across: (i) object- versus scene-centric, (ii) uniform versus long-tailed and (iii) general versus domain-specific datasets. Second, given the generality of the approach, we try to realize further gains with minor modifications. We show that learning additional invariances -- through the use of multi-scale cropping, stronger augmentations and nearest neighbors -- improves the representations. Finally, we observe that MoCo learns spatially structured representations when trained with a multi-crop strategy. The representations can be used for semantic segment retrieval and video instance segmentation without finetuning. Moreover, the results are on par with specialized models. We hope this work will serve as a useful study for other researchers. The code and models are available at https://github.com/wvangansbeke/Revisiting-Contrastive-SSL.

NeurIPS 2021. Code: https://github.com/wvangansbeke/Revisiting-Contrastive-SSL

References in corpus (4)

• Microsoft COCO Captions: Data Collection and Evaluation Server
• Unsupervised Deep Learning by Neighbourhood Discovery
• Unsupervised Learning of Dense Visual Representations
• Self-Supervised Visual Representation Learning from Hierarchical Grouping

Cited by in corpus (6)

• Supervision Exists Everywhere: A Data Efficient Contrastive Language-Image Pre-training Paradigm
• CMID: A Unified Self-Supervised Learning Framework for Remote Sensing Image Understanding
• Few-Shot Panoptic Segmentation With Foundation Models
• AdaTreeFormer: Few Shot Domain Adaptation for Tree Counting from a Single High-Resolution Image
• Self-Supervised Representation Learning for Nerve Fiber Distribution Patterns in 3D-PLI
• Object-Aware Cropping for Self-Supervised Learning