Search papers, labs, and topics across Lattice.
This study investigates how humans identify defects in AI-generated images produced by state-of-the-art text-to-image models, particularly when prompts involve complex compositional factors. By creating the compositional AI-generated image defect (CO-AID) dataset, which includes 651 reference images and detailed assessments from 29 participants, the authors reveal systematic flaws in the generated images. The findings indicate that training a deep learning model on the CO-AID dataset can effectively predict defects and enhance the quality of AI image generation.
Humans can pinpoint systematic flaws in AI-generated images, revealing critical insights into the limitations of current text-to-image models.
*Chulin Zhao and Ruoqi Hu contributed equally to this work. State-of-the-art text-to-image (T2I) models exhibit pronounced and systematic defects when prompts involve intricate compositional factors such as multiple entities and multiple attributes. In this paper, we investigate how humans identify such defects. Specifically, we manually select 651 reference images from the four categories of people, hand, object, and scene that exhibit complex compositional characteristics, from which prompts emphasizing compositional factors are derived by manually editing ChatGPT-generated prompts. We then feed the prompts into three selected T2I models to generate AI images and conduct a comprehensive subjective study to identify their defects. For each image, 29 participants provide multi-label assessments specifying defect types and locations. The study yields the compositional AI-generated image defect (CO-AID) dataset, including reference images, prompts, AI-generated images, and information on defect locations and types. Experimental results show that training a deep model on CO-AID can both predict defects in AI-generated images and optimize AI image generation, demonstrating its usability and effectiveness. The database and supplementary materials are available at: https://github.com/Future-IQA/CO-AID .
