Pith. sign in

REVIEW 3 cited by

Human Evaluation of Text-to-Image Models on a Multi-Task Benchmark

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2211.12112 v1 pith:PEJSR36Z submitted 2022-11-22 cs.CV cs.AIcs.LG

Human Evaluation of Text-to-Image Models on a Multi-Task Benchmark

classification cs.CV cs.AIcs.LG
keywords text-to-imagemodelsabilitybenchmarkpromptshumanmulti-taskobjects
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We provide a new multi-task benchmark for evaluating text-to-image models. We perform a human evaluation comparing the most common open-source (Stable Diffusion) and commercial (DALL-E 2) models. Twenty computer science AI graduate students evaluated the two models, on three tasks, at three difficulty levels, across ten prompts each, providing 3,600 ratings. Text-to-image generation has seen rapid progress to the point that many recent models have demonstrated their ability to create realistic high-resolution images for various prompts. However, current text-to-image methods and the broader body of research in vision-language understanding still struggle with intricate text prompts that contain many objects with multiple attributes and relationships. We introduce a new text-to-image benchmark that contains a suite of thirty-two tasks over multiple applications that capture a model's ability to handle different features of a text prompt. For example, asking a model to generate a varying number of the same object to measure its ability to count or providing a text prompt with several objects that each have a different attribute to identify its ability to match objects and attributes correctly. Rather than subjectively evaluating text-to-image results on a set of prompts, our new multi-task benchmark consists of challenge tasks at three difficulty levels (easy, medium, and hard) and human ratings for each generated image.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Video Deepfake Abuse: How Company Choices Predictably Shape Misuse Patterns

    cs.CY 2025-11 conditional novelty 6.0

    A few open-weight video models and distribution platforms dominate the creation and spread of NSFW AI video, making developer and platform choices the main intervention points for reducing non-consensual deepfake abuse.

  2. Understanding the Challenges and Opportunities of Generative AI Apps: An Empirical Study

    cs.SE 2025-06 unverdicted novelty 6.0

    Large-scale review mining of 1M+ comments from 171 Gen-AI apps using an LLM framework reveals top topics plus three opportunities and three challenges for developers.

  3. From LLM-Driven Trading Card Generation to Procedural Relatedness: A Pok\'emon Case Study

    cs.AI 2026-04 unverdicted novelty 5.0

    A pipeline combining LLMs, fine-tuned embeddings, and diffusion models generates personalized Pokémon cards; a 49-participant user study found high satisfaction and that most users realized their ideas via prompt tweaks.