Synthetic Curriculum Reinforces Compositional Text-to-Image Generation

TL;DR

This paper introduces CompGen, a reinforcement learning framework using scene graphs and curriculum strategies to improve the compositional capabilities of text-to-image models, especially in complex scene synthesis.

Contribution

It proposes a novel curriculum reinforcement learning method with adaptive graph sampling to enhance compositional T2I generation, integrating it into existing models.

Findings

CompGen improves compositional generation in diffusion and auto-regressive models.

Easy-to-hard and Gaussian curriculum strategies outperform random sampling.

Significant enhancement in complex scene synthesis capabilities.

Abstract

Text-to-Image (T2I) generation has long been an open problem, with compositional synthesis remaining particularly challenging. This task requires accurate rendering of complex scenes containing multiple objects that exhibit diverse attributes as well as intricate spatial and semantic relationships, demanding both precise object placement and coherent inter-object interactions. In this paper, we propose a novel compositional curriculum reinforcement learning framework named CompGen that addresses compositional weakness in existing T2I models. Specifically, we leverage scene graphs to establish a novel difficulty criterion for compositional ability and develop a corresponding adaptive Markov Chain Monte Carlo graph sampling algorithm. This difficulty-aware approach enables the synthesis of training curriculum data that progressively optimize T2I models through reinforcement learning. We integrate our curriculum learning approach into Group Relative Policy Optimization (GRPO) and investigate different curriculum scheduling strategies. Our experiments reveal that CompGen exhibits distinct scaling curves under different curriculum scheduling strategies, with easy-to-hard and Gaussian sampling strategies yielding superior scaling performance compared to random sampling. Extensive experiments demonstrate that CompGen significantly enhances compositional generation capabilities for both diffusion-based and auto-regressive T2I models, highlighting its effectiveness in improving the compositional T2I generation systems.