Interpretable method improves emotional editing of complex images
IECLA: Interpretable Emotion Contribution Learning for Affective Image Editing
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Changing a picture to show a specific emotion can be tricky because it's not clear which parts should change and which should stay the same. The authors created a new way to figure out which parts of the image contribute to the current feeling and which parts can be changed to show a new emotion. Their method helps editing tools make more precise changes, keeping the important details while expressing the desired emotion better. Tests and user feedback showed their method works well on different types of images.
What this means in practice
- •For digital artists: Use interpretable emotion contribution predictions to make targeted emotional enhancements in complex images while preserving key visual details.
- •For photo editing software developers: Integrate emotion contribution localization to guide emotion-driven editing operations that adaptively change image parts based on desired feelings.
Authors
Yijie Guo, Dexiang Hong, Weidong Chen, Xinyan Liu, Zhendong Mao
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.