CgT-GAN: CLIP-guided Text GAN for Image Captioning

TL;DR

This paper introduces CgT-GAN, a novel image captioning model that leverages CLIP guidance and adversarial training to generate more natural and semantically accurate captions without relying on human-annotated datasets.

Contribution

The paper proposes CgT-GAN, integrating images into training with adversarial and CLIP-based semantic rewards, improving captioning quality without human annotations.

Findings

Outperforms state-of-the-art methods on three captioning subtasks.

Uses novel CLIP-agg semantic guidance for better caption alignment.

Demonstrates significant improvements in caption naturalness and semantic accuracy.

Abstract

The large-scale visual-language pre-trained model, Contrastive Language-Image Pre-training (CLIP), has significantly improved image captioning for scenarios without human-annotated image-caption pairs. Recent advanced CLIP-based image captioning without human annotations follows a text-only training paradigm, i.e., reconstructing text from shared embedding space. Nevertheless, these approaches are limited by the training/inference gap or huge storage requirements for text embeddings. Given that it is trivial to obtain images in the real world, we propose CLIP-guided text GAN (CgT-GAN), which incorporates images into the training process to enable the model to "see" real visual modality. Particularly, we use adversarial training to teach CgT-GAN to mimic the phrases of an external text corpus and CLIP-based reward to provide semantic guidance. The caption generator is jointly rewarded based on the caption naturalness to human language calculated from the GAN's discriminator and the semantic guidance reward computed by the CLIP-based reward module. In addition to the cosine similarity as the semantic guidance reward (i.e., CLIP-cos), we further introduce a novel semantic guidance reward called CLIP-agg, which aligns the generated caption with a weighted text embedding by attentively aggregating the entire corpus. Experimental results on three subtasks (ZS-IC, In-UIC and Cross-UIC) show that CgT-GAN outperforms state-of-the-art methods significantly across all metrics. Code is available at https://github.com/Lihr747/CgtGAN.