Data-Efficient Language-Supervised Zero-Shot Learning with Self-Distillation
This work addresses the problem of reducing data requirements for training zero-shot vision-language models, which is incremental as it builds on existing methods like CLIP but offers significant efficiency gains.
The paper tackles the data inefficiency of language-supervised zero-shot learning by proposing a contrastive distillation method that uses soft labels from noisy image-text pairs, achieving strong performance with only 3M pairs (133x smaller than CLIP) and exceeding previous state-of-the-art by 73% on ImageNet 21k+1k and beating CLIP by 10.5% on Google Open Images.
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).