Data-Efficient Language-Supervised Zero-Shot Learning with Self-Distillation
arXiv:2104.08945
Abstract
Traditional computer vision models are trained to predict a fixed set of predefined categories. Recently, natural language has been shown to be a broader and richer source of supervision that provides finer descriptions to visual concepts than supervised "gold" labels. Previous works, such as CLIP, use a simple pretraining task of predicting the pairings between images and text captions. CLIP, however, is data hungry and requires more than 400M image text pairs for training. We propose a data-efficient contrastive distillation method that uses soft labels to learn from noisy image-text pairs. Our model transfers knowledge from pretrained image and sentence encoders and achieves strong performance with only 3M image text pairs, 133x smaller than CLIP. Our method exceeds the previous SoTA of general zero-shot learning on ImageNet 21k+1k by 73% relatively with a ResNet50 image encoder and DeCLUTR text encoder. We also beat CLIP by 10.5% relatively on zero-shot evaluation on Google Open Images (19,958 classes).
arXiv comment4 pages, 1 figure
References in corpus (10)
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Distilling the Knowledge in a Neural Network
- A Simple Framework for Contrastive Learning of Visual Representations
- Learning Transferable Visual Models From Natural Language Supervision
- Language Models are Few-Shot Learners
- Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision
- A large annotated corpus for learning natural language inference
- Contrastive Learning of Medical Visual Representations from Paired Images and Text
- Label Refinery: Improving ImageNet Classification through Label Progression
- DeCLUTR: Deep Contrastive Learning for Unsupervised Textual Representations