CatLIP: CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data

TL;DR

This paper introduces CatLIP, a faster pre-training method for vision models on web-scale image-text data that reframes contrastive learning as a classification task, significantly accelerating training while maintaining high accuracy.

Contribution

It proposes a novel weakly supervised pre-training approach that eliminates pairwise similarity computations, achieving 2.7x faster training without sacrificing performance.

Findings

Achieves 2.7x faster training speed compared to contrastive learning.

Maintains high representation quality across diverse vision tasks.

Provides open-source code and pre-trained models.

Abstract

Contrastive learning has emerged as a transformative method for learning effective visual representations through the alignment of image and text embeddings. However, pairwise similarity computation in contrastive loss between image and text pairs poses computational challenges. This paper presents a novel weakly supervised pre-training of vision models on web-scale image-text data. The proposed method reframes pre-training on image-text data as a classification task. Consequently, it eliminates the need for pairwise similarity computations in contrastive loss, achieving a remarkable $2.7\times$ acceleration in training speed compared to contrastive learning on web-scale data. Through extensive experiments spanning diverse vision tasks, including detection and segmentation, we demonstrate that the proposed method maintains high representation quality. Our source code along with pre-trained model weights and training recipes is available at \url{https://github.com/apple/corenet}.