EmoKGEdit: Training-free Affective Injection via Visual Cue Transformation

TL;DR

This work tackles the challenge of controllable image emotion editing by introducing EmoKGEdit, a training-free framework that grounds emotional manipulation in a Multimodal Sentiment Association Knowledge Graph (MSA-KG). The system combines an Emotion Region-aware Module to localize affective loci, an Emotion Cue Transfer module guided by Chain-of-Thought reasoning, and a Disentangled Structure-Emotion Editing module to inject emotion within localized regions while preserving layout. By grounding edits in explicit object-scene-emotion relations and enforcing a dual-stream diffusion process, EmoKGEdit achieves superior performance in emotion fidelity and content preservation, as demonstrated by quantitative metrics and user studies. The approach offers interpretable, region-level emotion editing with practical implications for generating emotionally coherent imagery without compromising semantic structure.

Abstract

Existing image emotion editing methods struggle to disentangle emotional cues from latent content representations, often yielding weak emotional expression and distorted visual structures. To bridge this gap, we propose EmoKGEdit, a novel training-free framework for precise and structure-preserving image emotion editing. Specifically, we construct a Multimodal Sentiment Association Knowledge Graph (MSA-KG) to disentangle the intricate relationships among objects, scenes, attributes, visual clues and emotion. MSA-KG explicitly encode the causal chain among object-attribute-emotion, and as external knowledge to support chain of thought reasoning, guiding the multimodal large model to infer plausible emotion-related visual cues and generate coherent instructions. In addition, based on MSA-KG, we design a disentangled structure-emotion editing module that explicitly separates emotional attributes from layout features within the latent space, which ensures that the target emotion is effectively injected while strictly maintaining visual spatial coherence. Extensive experiments demonstrate that EmoKGEdit achieves excellent performance in both emotion fidelity and content preservation, and outperforms the state-of-the-art methods.

Figures (12)

• Figure 1: Visual comparison. While Emoedit and EmoEditor suffer from over-editing and under-editing respectively, our method successfully alters the emotional tone while maintaining high visual realism and semantic fidelity.
• Figure 2: Inspired by cognitive psychology on visual emotion processing, extracting emotion-inducing regions as scene cores and coupling them with their corresponding objects. Through this region–object coupling, the model focuses on the most emotionally salient content.
• Figure 3: Overview of EmoKGEdit. The proposed framework comprises four components: Multimodal Sentiment Association Knowledge Graph(MSA Knowledge Graph), Emotion Region-aware Module, Emotion Cue Transfer Module, and Disentangled Structure--Emotion Editing(DSEE) Module.
• Figure 4: Schematic diagram of MSA-KG
• Figure 5: Emotion Cue Transfer(ECT) with Chain-of-Thought Reasoning.