Evaluating Hallucination in Text-to-Image Diffusion Models with Scene-Graph based Question-Answering Agent

• URL: http://arxiv.org/abs/2412.05722v1
• Date: Sat, 07 Dec 2024 18:44:38 GMT
• Title: Evaluating Hallucination in Text-to-Image Diffusion Models with Scene-Graph based Question-Answering Agent
• Authors: Ziyuan Qin, Dongjie Cheng, Haoyu Wang, Huahui Yi, Yuting Shao, Zhiyuan Fan, Kang Li, Qicheng Lao,
• Abstract summary: An effective Text-to-Image (T2I) evaluation metric should accomplish the following: detect instances where the generated images do not align with the textual prompts. We propose a method based on large language models (LLMs) for conducting question-answering with an extracted scene-graph and created a dataset with human-rated scores for generated images.
• Score: 9.748808189341526
• License:
• Abstract: Contemporary Text-to-Image (T2I) models frequently depend on qualitative human evaluations to assess the consistency between synthesized images and the text prompts. There is a demand for quantitative and automatic evaluation tools, given that human evaluation lacks reproducibility. We believe that an effective T2I evaluation metric should accomplish the following: detect instances where the generated images do not align with the textual prompts, a discrepancy we define as the `hallucination problem' in T2I tasks; record the types and frequency of hallucination issues, aiding users in understanding the causes of errors; and provide a comprehensive and intuitive scoring that close to human standard. To achieve these objectives, we propose a method based on large language models (LLMs) for conducting question-answering with an extracted scene-graph and created a dataset with human-rated scores for generated images. From the methodology perspective, we combine knowledge-enhanced question-answering tasks with image evaluation tasks, making the evaluation metrics more controllable and easier to interpret. For the contribution on the dataset side, we generated 12,000 synthesized images based on 1,000 composited prompts using three advanced T2I models. Subsequently, we conduct human scoring on all synthesized images and prompt pairs to validate the accuracy and effectiveness of our method as an evaluation metric. All generated images and the human-labeled scores will be made publicly available in the future to facilitate ongoing research on this crucial issue. Extensive experiments show that our method aligns more closely with human scoring patterns than other evaluation metrics.

Related papers

• Human Body Restoration with One-Step Diffusion Model and A New Benchmark [74.66514054623669]
  We propose a high-quality dataset automated cropping and filtering (HQ-ACF) pipeline. This pipeline leverages existing object detection datasets and other unlabeled images to automatically crop and filter high-quality human images. We also propose emphOSDHuman, a novel one-step diffusion model for human body restoration.
  arXiv  Detail & Related papers  (2025-02-03T14:48:40Z)
• ANID: How Far Are We? Evaluating the Discrepancies Between AI-synthesized Images and Natural Images through Multimodal Guidance [19.760989919485894]
  We introduce an AI-Natural Image Discrepancy Evaluation benchmark aimed at addressing the critical question: textithow far are AI-generated images from truly realistic images? We have constructed a large-scale multimodal dataset, the Distinguishing Natural and AI-generated Images (DNAI) dataset, which includes over 440,000 AIGI samples generated by 8 representative models. Our fine-grained assessment framework provides a comprehensive evaluation of the DNAI dataset across five key dimensions.
  arXiv  Detail & Related papers  (2024-12-23T15:08:08Z)
• Image2Text2Image: A Novel Framework for Label-Free Evaluation of Image-to-Text Generation with Text-to-Image Diffusion Models [16.00576040281808]
  We propose a novel framework called Image2Text2Image to evaluate image captioning models. A high similarity score suggests that the model has produced a faithful textual description, while a low score highlights discrepancies. Our framework does not rely on human-annotated captions reference, making it a valuable tool for assessing image captioning models.
  arXiv  Detail & Related papers  (2024-11-08T17:07:01Z)
• Evaluating Multiview Object Consistency in Humans and Image Models [68.36073530804296]
  We leverage an experimental design from the cognitive sciences which requires zero-shot visual inferences about object shape. We collect 35K trials of behavioral data from over 500 participants. We then evaluate the performance of common vision models.
  arXiv  Detail & Related papers  (2024-09-09T17:59:13Z)
• A Novel Evaluation Framework for Image2Text Generation [15.10524860121122]
  We propose an evaluation framework rooted in a modern large language model (LLM) capable of image generation. A high similarity score suggests that the image captioning model has accurately generated textual descriptions. A low similarity score indicates discrepancies, revealing potential shortcomings in the model's performance.
  arXiv  Detail & Related papers  (2024-08-03T09:27:57Z)
• Multi-Modal Prompt Learning on Blind Image Quality Assessment [65.0676908930946]
  Image Quality Assessment (IQA) models benefit significantly from semantic information, which allows them to treat different types of objects distinctly. Traditional methods, hindered by a lack of sufficiently annotated data, have employed the CLIP image-text pretraining model as their backbone to gain semantic awareness. Recent approaches have attempted to address this mismatch using prompt technology, but these solutions have shortcomings. This paper introduces an innovative multi-modal prompt-based methodology for IQA.
  arXiv  Detail & Related papers  (2024-04-23T11:45:32Z)
• Stellar: Systematic Evaluation of Human-Centric Personalized Text-to-Image Methods [52.806258774051216]
  We focus on text-to-image systems that input a single image of an individual and ground the generation process along with text describing the desired visual context. We introduce a standardized dataset (Stellar) that contains personalized prompts coupled with images of individuals that is an order of magnitude larger than existing relevant datasets and where rich semantic ground-truth annotations are readily available. We derive a simple yet efficient, personalized text-to-image baseline that does not require test-time fine-tuning for each subject and which sets quantitatively and in human trials a new SoTA.
  arXiv  Detail & Related papers  (2023-12-11T04:47:39Z)
• Let's ViCE! Mimicking Human Cognitive Behavior in Image Generation Evaluation [96.74302670358145]
  We introduce an automated method for Visual Concept Evaluation (ViCE) to assess consistency between a generated/edited image and the corresponding prompt/instructions. ViCE combines the strengths of Large Language Models (LLMs) and Visual Question Answering (VQA) into a unified pipeline, aiming to replicate the human cognitive process in quality assessment.
  arXiv  Detail & Related papers  (2023-07-18T16:33:30Z)
• ArtWhisperer: A Dataset for Characterizing Human-AI Interactions in Artistic Creations [26.4215586218117]
  This work investigates how people use text-to-image models to generate desired target images. We created ArtWhisperer, an online game where users are given a target image and are tasked with iteratively finding a prompt that creates a similar-looking image as the target. We recorded over 50,000 human-AI interactions; each interaction corresponds to one text prompt created by a user and the corresponding generated image.
  arXiv  Detail & Related papers  (2023-06-13T21:10:45Z)
• TeTIm-Eval: a novel curated evaluation data set for comparing text-to-image models [1.1252184947601962]
  evaluating and comparing text-to-image models is a challenging problem. In this paper a novel evaluation approach is experimented, on the basis of: (i) a curated data set, divided into ten categories; (ii) a quantitative metric, the CLIP-score, (iii) a human evaluation task to distinguish, for a given text, the real and the generated images. Early experimental results show that the accuracy of the human judgement is fully coherent with the CLIP-score.
  arXiv  Detail & Related papers  (2022-12-15T13:52:03Z)
• Transparent Human Evaluation for Image Captioning [70.03979566548823]
  We develop a rubric-based human evaluation protocol for image captioning models. We show that human-generated captions show substantially higher quality than machine-generated ones. We hope that this work will promote a more transparent evaluation protocol for image captioning.
  arXiv  Detail & Related papers  (2021-11-17T07:09:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.