cs.CVOct 29, 2026

Grounding Image Composition with Label-Conditioned Visual Evidence

Authors: Armin Dadras, Tingyu Lin, Robert Sablatnig, Franziska Proksa, Markus Seidl

Organizations: Computer Vision Lab, TU Wien, Vienna, Austria · UAS St. Pölten, St. Pölten, Austria

Abstract

Image composition labels such as center, diagonal, or rule of thirds describe visual structure rather than object categories. KU-PCP provides image-level labels and class-specific geometric elements, but does not show which visible regions people associate with each of its nine classes. We introduce KU-PCP Elements (KU-PCPE), a label-conditioned region-evidence benchmark for photographic composition. Given an image and a label already judged applicable, the task is to segment the visible regions that explain why the label applies. KU-PCPE contains 4,093 single-annotator training masks and a 1,327-task expert test split with three masks per image-label task. The annotations also include exploratory segment-versus-line ratings that describe whether region or semantic-guideline evidence better explains the label. A held-out expert reaches .867 Dice against the other experts, while the strongest deployable model reaches .664, a proposal oracle .672, and the best vision-language model (VLM) proposal selector .395. KU-PCPE exposes the gap between recognizing a composition label and grounding the human-aligned visual evidence that makes it plausible in a particular image.

Figures & tables

Appendix figures & tables41 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. CompositionEvidence: Controlling Geometric Evidence for Cinematic Composition Reasoning with Frozen Vision-Language Models

    Nov 7, 2026Yichen He, Xin Huang, Yuxi LiuCompositional Reasoning in VLMsVisual Evidence Sensitivity in VLMs

  2. COMPASS: Grounding Composition-Intent Guidance in Unified Multimodal Models

    Jun 27, 2026Ziqi Zhou, Weize Quan, Mining Tan +6Multimodal GroundingUnified Multimodal Models