aiqbench analysis
This paper proposes a method for editing images to convey a target emotion while preserving content. It first learns which source regions support or oppose the current emotion, using counterfactual interventions, and then drives existing editors with that signal. The reported gains are measured across six editing backbones, but they rely on benchmark evaluations, VLM judges, and user studies, so practical value depends on how well those results hold outside the tested datasets.
IECLA: Interpretable Emotion Contribution Learning for Affective Image Editing
arXiv:2610.25015v1 Announce Type: cross Abstract: Affective image editing (AIE) aims to evoke a target emotion while preserving the content and structure of a source image. Existing methods improve emotional expression or visual fidelity, but the target emotion alone does not reveal which source regions should be changed and which should remain intact, especially in complex scenes. We introduce Interpretable Emotion Contribution Learning for Affective Image Editing (IECLA), a framework that separates source-emotion attribution from target-aware editing. IECLA constructs signed object-level supervision through counterfactual interventions and trains a context-conditioned contribution predictor to localize evidence that supports or opposes the source emotion. At inference, an image-level evaluator estimates the source emotion directly from the input, while the predictor identifies editable and preservation evidence without repeated interventions. A target-aware editing agent then translates this evidence and the desired emotion into localized, content-aware operations for existing editors. Extensive evaluations show that IECLA predicts counterfactual contributions more faithfully on both in-domain and cross-dataset tests, while external VLM and independent human references support the quality of its localized evidence. Across six editing backbones, IECLA consistently strengthens target-emotion expression while preserving image content and structure, and user studies further show improved perceived target expression and naturalness.
Abstract:Affective image editing (AIE) aims to evoke a target emotion while preserving the content and structure of a source image. Existing methods improve emotional expression or visual fidelity, but the target emotion alone does not reveal which source regions should be changed and which should remain intact, especially in complex scenes. We introduce Interpretable Emotion Contribution Learning for Affective Image Editing (IECLA), a framework that separates source-emotion attribution from target-aware editing. IECLA constructs signed object-level supervision through counterfactual interventions and trains a context-conditioned contribution predictor to localize evidence that supports or opposes the source emotion. At inference, an image-level evaluator estimates the source emotion directly from the input, while the predictor identifies editable and preservation evidence without repeated interventions. A target-aware editing agent then translates this evidence and the desired emotion into localized, content-aware operations for existing editors. Extensive evaluations show that IECLA predicts counterfactual contributions more faithfully on both in-domain and cross-dataset tests, while external VLM and independent human references support the quality of its localized evidence. Across six editing backbones, IECLA consistently strengthens target-emotion expression while preserving image content and structure, and user studies further show improved perceived target expression and naturalness.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2610.25015 [cs.CV]
(or arXiv:2610.25015v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.25015 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 292 words
Related coverage
Subject-Specific Analysis of Self-Initiated Attention Shifts from EEG with Controlled Internal and External Attention Conditions
arXiv · Sep 16, 2026
Twitch Plays: AI Interpretability of Bonsai-8B
LessWrong · Oct 19, 2026
Interpretable by Design Tabular Prediction via In-Context Local Linear Learning
arXiv · Oct 30, 2026
Beyond Gate Weights: Decision-Level Interpretation of Radiomics-Image Feature Fusion for RCC Classification
arXiv · Oct 15, 2026