Text-to-Image Diffusion Models are Zero-Shot Classifiers
The excellent generative capabilities of text-to-image diffusion models suggest they learn informative representations of image-text data. However, what knowledge their representations capture is not fully understood, and they have not been thoroughly explored on downstream tasks. We investigate diffusion models by proposing a method for evaluating them as zero-shot classifiers. The key idea is using a diffusion model's ability to denoise a noised image given a text description of a label as a proxy for that label's likelihood. We apply our method to Stable Diffusion and Imagen, using it to probe fine-grained aspects of the models' knowledge and comparing them with CLIP's zero-shot abilities. They perform competitively with CLIP on a wide range of zero-shot image classification datasets. Additionally, they achieve state-of-the-art results on shape/texture bias tests and can successfully perform attribute binding while CLIP cannot. Although generative pre-training is prevalent in NLP, visual foundation models often use other methods such as contrastive learning. Based on our findings, we argue that generative pre-training should be explored as a compelling alternative for vision-language tasks.
Code (1)
Tasks
AttributeContrastive Learningimage-classificationImage ClassificationZero-Shot Image ClassificationMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
A Simple and Efficient Baseline for Zero-Shot Generative Classification
Large diffusion models have become mainstream generative models in both academic studies and industrial AIGC applications. Recently, a number of works further explored how to employ the power of large diffusion models as…
zero-shot-classificationZero-Shot LearningText-to-Image Diffusion Models are Zero Shot Classifiers
The excellent generative capabilities of text-to-image diffusion models suggest they learn informative representations of image-text data. However, what knowledge their representations capture is not fully understood, an…
Your Diffusion Model is Secretly a Zero-Shot Classifier
The recent wave of large-scale text-to-image diffusion models has dramatically increased our text-based image generation abilities. These models can generate realistic images for a staggering variety of prompts and exhib…
Domain GeneralizationFine-Grained Image ClassificationImage ClassificationImage Generation+5Self-Improving Diffusion Classifiers with Minority Preference Optimization
Prior studies have demonstrated that diffusion classifiers achieve robust zero-shot classification performance. However, their effectiveness is strongly tied to the pretraining data distribution: they perform well in maj…
How Do Diffusion Classifiers Decide? A Bias-Centric Evaluation
Diffusion models have recently been repurposed for zero-shot classification, giving rise to diffusion classifiers that identify the best-matching text prompt by minimizing the noise-prediction error. Despite their growin…