Awesome Vision-Language Models
The companion repository for a systematic academic survey of vision-language models for visual recognition tasks, curating papers by pre-training method, transfer learning and application.
Awesome Vision-Language Models is the repository accompanying the paper Vision-Language Models for Vision Tasks: A Survey, published in IEEE Transactions on Pattern Analysis and Machine Intelligence in 2024 and selected for the journal's Top 50 Popular Paper List. It surveys vision-language model studies across visual recognition tasks such as image classification, object detection and semantic segmentation.
It organizes curated papers into sections covering synthetic data for vision-language models, pre-training methods and transfer-learning methods, with each entry linking to its paper and, where available, its code or project page. It also points to a companion collection on agentic multimodal large language models. Contributions of missing papers are accepted through pull requests that add a title, paper link, conference and code or project link in the documented table format.
It contains no dedicated code of its own beyond this curated reference list.