Pith. sign in

REVIEW 4 cited by

HumanAesExpert: Advancing a Multi-Modality Foundation Model for Human Image Aesthetic Assessment

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 2503.23907 v2 pith:CTXBW7A6 submitted 2025-03-31 cs.CV cs.AI

HumanAesExpert: Advancing a Multi-Modality Foundation Model for Human Image Aesthetic Assessment

classification cs.CV cs.AI
keywords aesthetichiaahumanhumanaesexpertassessmentimageimagesmodel
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Image Aesthetic Assessment (IAA) is a long-standing and challenging research task. However, its subset, Human Image Aesthetic Assessment (HIAA), has been scarcely explored. To bridge this research gap, our work pioneers a holistic implementation framework tailored for HIAA. Specifically, we introduce HumanBeauty, the first dataset purpose-built for HIAA, which comprises 108k high-quality human images with manual annotations. To achieve comprehensive and fine-grained HIAA, 50K human images are manually collected through a rigorous curation process and annotated leveraging our trailblazing 12-dimensional aesthetic standard, while the remaining 58K with overall aesthetic labels are systematically filtered from public datasets. Based on the HumanBeauty database, we propose HumanAesExpert, a powerful Vision Language Model for aesthetic evaluation of human images. We innovatively design an Expert head to incorporate human knowledge of aesthetic sub-dimensions while jointly utilizing the Language Modeling (LM) and Regression heads. This approach empowers our model to achieve superior proficiency in both overall and fine-grained HIAA. Furthermore, we introduce a MetaVoter, which aggregates scores from all three heads, to effectively balance the capabilities of each head, thereby realizing improved assessment precision. Extensive experiments demonstrate that our HumanAesExpert models deliver significantly better performance in HIAA than other state-of-the-art models. Project webpage: https://humanaesexpert.github.io/HumanAesExpert/

discussion (0)

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

Forward citations

Cited by 4 Pith papers

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

  1. MultiEmo-Bench: Multi-label Visual Emotion Analysis for Multi-modal Large Language Models

    cs.CV 2026-05 conditional novelty 7.0

    MultiEmo-Bench supplies 10,344 images with aggregated multi-label emotion votes from 20 annotators each to evaluate MLLMs on dominant emotion and full distribution prediction.

  2. PortraitGen: Exemplar-Driven GRPO with Dual-Reward Guidance for Photorealistic Portrait Generation

    cs.CV 2026-06 unverdicted novelty 6.0

    PortraitGen integrates real-image exemplars into GRPO sampling and applies dual rewards (OmniReward and AI-Portrait) to improve photorealism, claiming better results than baselines on a new PortraitBench.

  3. CoInteract: Physically-Consistent Human-Object Interaction Video Synthesis via Spatially-Structured Co-Generation

    cs.CV 2026-04 unverdicted novelty 6.0

    CoInteract adds a human-aware mixture-of-experts and spatially-structured co-generation to a diffusion transformer to synthesize videos with stable structures and physically plausible human-object contacts.

  4. PureCC: Pure Learning for Text-to-Image Concept Customization

    cs.CV 2026-03 unverdicted novelty 5.0

    PureCC introduces a decoupled learning objective, dual-branch training pipeline with frozen extractor, and adaptive guidance scale λ* for high-fidelity concept customization while preserving original model behavior in...