Research map / Detection and segmentation / CNN features
SAFL-Net: Semantic-Agnostic Feature Learning Network with Auxiliary Plugins for Image Manipulation Detection
Zhihao Sun, Haoran Jiang, Danding Wang, Xirong Li, Juan Cao
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 dedicated classification head instead of averaging pixel outputs (ablation) | SAFL-Net baseline · Coverage | pixel-level AUC: 74.1 -> 81.6 | Improved |
| add plain boundary prediction auxiliary task (ablation) | baseline · IMD20 | pixel-level AUC: 84.6 -> 79.7 | Hurt |
| replace plain boundary task with Boundary Guidance Module differential block (ablation) | baseline · IMD20 | pixel-level AUC: 79.7 -> 87.3 | Improved |
| add soft boundary supervision and tailored BGM loss (ablation) | BGM · IMD20 | pixel-level AUC: 87.3 -> 89.9 | Improved |
| add semantic suppression image-level constraint (ablation) | baseline + BGM · IMD20 | pixel-level AUC: 89.9 -> 93.1 | Improved |
| add patch-level contrastive loss in SSM (ablation) | baseline + BGM + SSM#1 · CASIA | pixel-level AUC: 86.1 -> 90.9 | Improved |
| add patch-level contrastive loss in SSM (ablation) | baseline + BGM + SSM#1 · Coverage | pixel-level AUC: 93.8 -> 93.5 | No clear effect |
| SAFL-Net versus PSCC-Net on Columbia | SAFL-Net · Columbia pretrained | pixel-level AUC: 98.2 vs 96.9 | Hurt |
| SAFL-Net versus ObjectFormer on image-level AUC | SAFL-Net · CASIA-D | image-level AUC: 99.70 vs 99.48 | Hurt |
Later papers that compared with this method
1 later accepted paper reports SAFL-Net as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Noise-assisted Prompt Learning for Image Forgery Detection and Localization (ECCV 2024 · better on 10 of 10 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- UnionFormer: Unified-Learning Transformer with Multi-View Representation for Image Manipulation Detection and Localization (CVPR 2024 · similarity 0.84)
- Learning to Manipulate Artistic Images (AAAI 2024 · similarity 0.84)
- A New Benchmark and Model for Challenging Image Manipulation Detection (AAAI 2024 · similarity 0.84)
- ForgerySleuth: Empowering Multimodal Large Language Models for Image Manipulation Detection (NeurIPS 2025 · similarity 0.83)
- IDseq: Decoupled and Sequentially Detecting and Grounding Multi-Modal Media Manipulation (AAAI 2025 · similarity 0.83)
- Unleashing In-context Learning of Autoregressive Models for Few-shot Image Manipulation (CVPR 2025 · similarity 0.83)
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.
