SILC: Improving Vision Language Pretraining with Self-Distillation


Loading...

Date

2023-10-20

Publication Type

Working Paper

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

Image-Text pretraining on web-scale image caption dataset has become the default recipe for open vocabulary classification and retrieval models thanks to the success of CLIP and its variants. Several works have also used CLIP features for dense prediction tasks and have shown the emergence of open-set abilities. However, the contrastive objective only focuses on image-text alignment and does not incentivise image feature learning for dense prediction tasks. In this work, we propose the simple addition of local-to-global correspondence learning by self-distillation as an additional objective for contrastive pre-training to propose SILC. We show that distilling local image features from an exponential moving average (EMA) teacher model significantly improves model performance on several computer vision tasks including classification, retrieval, and especially segmentation. We further show that SILC scales better with the same training duration compared to the baselines. Our model SILC sets a new state of the art for zero-shot classification, few shot classification, image and text retrieval, zero-shot segmentation, and open vocabulary segmentation.

Publication status

published

Editor

Book title

Journal / series

Volume

Pages / Article No.

2310.13355

Publisher

Cornell University

Event

Edition / version

v1

Methods

Software

Geographic location

Date collected

Date created

Subject

Organisational unit

03514 - Van Gool, Luc (emeritus) / Van Gool, Luc (emeritus) check_circle

Notes

Funding

Related publications and datasets