SVL-Adapter: Self-Supervised Adapter for Vision-Language Pretrained Models

Omiros Pantazis, Gabriel Brostow, Kate E. Jones, Oisin Mac Aodha

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

Vision-language models such as CLIP are pretrained on large volumes of internet sourced image and text pairs, and have been shown to sometimes exhibit impressive zero- and low-shot image classification performance. However, due to their size, fine-tuning these models on new datasets can be prohibitively expensive, both in terms of the supervision and compute required. To combat this, a series of light-weight adaptation methods have been proposed to efficiently adapt such models when limited supervision is available. In this work, we show that while effective on internet-style datasets, even those remedies under-deliver on classification tasks with images that differ significantly from those commonly found online. To address this issue, we present a new approach called SVL-Adapter that combines the complementary strengths of both vision-language pretraining and self-supervised representation learning. We report an average classification accuracy improvement of 10% in the low-shot setting when compared to existing methods, on a set of challenging visual classification tasks. Further, we present a fully automatic way of selecting an important blending hyperparameter for our model that does not require any held-out labeled validation data.
Code for our project is available here:
Original languageEnglish
Title of host publicationProceedings of The 33rd British Machine Vision Conference (BMVC 2022)
PublisherBMVA Press
Number of pages23
Publication statusPublished - 25 Nov 2022
EventThe 33rd British Machine Vision Conference, 2022 - London, United Kingdom
Duration: 21 Nov 202224 Nov 2022
Conference number: 33


ConferenceThe 33rd British Machine Vision Conference, 2022
Abbreviated titleBMVC 2022
Country/TerritoryUnited Kingdom
Internet address


Dive into the research topics of 'SVL-Adapter: Self-Supervised Adapter for Vision-Language Pretrained Models'. Together they form a unique fingerprint.

Cite this