CLIP-Diffusion-LM: Apply Diffusion Model on Image Captioning
This addresses image captioning for vision-language tasks, but it is incremental as it adapts an existing diffusion method to a new domain.
The paper tackles image captioning by applying a denoising diffusion model to text generation, achieving a BLEU-4 score of 0.1876 on Flickr8k and 0.2470 on a combined dataset with fewer inference steps than autoregressive models.
Image captioning task has been extensively researched by previous work. However, limited experiments focus on generating captions based on non-autoregressive text decoder. Inspired by the recent success of the denoising diffusion model on image synthesis tasks, we apply denoising diffusion probabilistic models to text generation in image captioning tasks. We show that our CLIP-Diffusion-LM is capable of generating image captions using significantly fewer inference steps than autoregressive models. On the Flickr8k dataset, the model achieves 0.1876 BLEU-4 score. By training on the combined Flickr8k and Flickr30k dataset, our model achieves 0.2470 BLEU-4 score. Our code is available at https://github.com/xu-shitong/diffusion-image-captioning.