Research map / Image generation / Editing and style transfer
Inter-Edit: First Benchmark for Interactive Instruction-Based Image Editing
Delong Liu, Haotian Hou, Zhaohui Hou, Zhiyuan Huang, Shihao Han, Mingjie Zhan, Zhicheng Zhao, Fei Su
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MLLM filtering of synthetic training data (ablation) | Inter-Edit RNI · Inter-Edit test set | S_VQA: 6.189 -> 6.431 | Improved |
| add MLLM filtering of synthetic training data (ablation) | Inter-Edit CJT · Inter-Edit test set | S_VQA: 6.087 -> 6.333 | Improved |
| add MLLM filtering of synthetic training data (ablation) | Inter-Edit CIA · Inter-Edit test set | S_VQA: 5.792 -> 5.979 | Improved |
| add mask post-processing (morphology, blur) to simulate user masks (ablation) | Inter-Edit RNI · Inter-Edit test set | BDS: 12.239 -> 10.485 | Improved |
| add mask post-processing (ablation) | Inter-Edit CJT · Inter-Edit test set | S_VQA: 6.106 -> 6.333 | Improved |
| fine-tune Q-Edit with LoRA on Inter-Edit (ablation) | Inter-Edit CIA · Inter-Edit test set | S_VQA: 5.477 -> 5.979 | Improved |
| lower LoRA rank from 32 to 16 (ablation) | Inter-Edit CJT · Inter-Edit test set | S_VQA: 6.333 -> 6.203 | Hurt |
| raise LoRA rank from 32 to 64 (ablation) | Inter-Edit CJT · Inter-Edit test set | S_VQA: 6.333 -> 6.346 | No clear effect |
| raise LoRA rank from 32 to 64 (ablation) | Inter-Edit CIA · Inter-Edit test set | S_VQA: 5.979 -> 5.964 | No clear effect |
| use mask plus concise instruction (I3E RNI) instead of instruction-only Q-Edit | Inter-Edit RNI · Inter-Edit test set | S_VQA: 5.695 -> 6.431 | Improved |
| use mask-based inpainting (Flux-Fill) instead of I3E | Flux-Fill · Inter-Edit test set | BDS: 15.969 vs 10.485 | Hurt |
Closest papers by meaning
- Balancing Preservation and Modification: A Region and Semantic Aware Metric for Instruction-Based Image Editing (ICML 2025 · similarity 0.90)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.90)
- I2EBench: A Comprehensive Benchmark for Instruction-based Image Editing (NeurIPS 2024 · similarity 0.89)
- Text-Vision Co-Instructed Image Editing (NeurIPS 2026 · similarity 0.89)
- InsightEdit: Towards Better Instruction Following for Image Editing (CVPR 2025 · similarity 0.89)
- Learning Action and Reasoning-Centric Image Editing from Videos and Simulation (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
