Conditional Prompt Learning for Vision-Language Models

TLDR

Pre‑trained vision‑language models such as CLIP require adaptation to downstream datasets, and the recently proposed Context Optimization (CoOp) adapts them by learning prompt tokens. This work proposes Conditional Context Optimization (CoCoOp) to overcome CoOp’s overfitting to base classes by generating image‑conditional tokens. CoCoOp extends CoOp by training a lightweight neural network that produces a dynamic prompt token conditioned on each image, replacing the static prompt vectors. CoCoOp achieves substantially better generalization to unseen classes and improved domain generalization compared to CoOp, as shown by extensive experiments. Code is available at https://github.com/KaiyangZhou/CoOp.

Abstract

With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning—a recent trend in NLP—to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at https://github.com/KaiyangZhou/CoOp.