Research map / Detection and segmentation / CNN features
ManTra-Net: Manipulation Tracing Network for Detection and Localization of Image Forgeries With Anomalous Features
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use ManTra-Net without dataset finetuning | ManTra-Net · COVERAGE | pixel-level AUC: 81.7% -> 81.9% | Helped |
| Use ManTra-Net without dataset finetuning | ManTra-Net · CASIA | pixel-level AUC: 79.5% -> 81.7% | Helped |
| Use ManTra-Net instead of RGB-N | ManTra-Net · NIST 2016 | pixel-level AUC: 93.7% -> 79.5% | Hurt |
| Use ManTra-Net instead of RGB-N | ManTra-Net · Columbia | pixel-level AUC: 85.8% -> 82.4% | Hurt |
| Use a wider and deeper backbone (ablation) | IMC-VGG · IMC-385 | Top-1 accuracy: 47.3% -> 51.8% | Helped |
| Increase manipulation-class granularity from 7 to 385 classes (ablation) | IMC feature learning · KCMI | accuracy: 57.2% -> 82.6% | Helped |
| Combine Conv2D, BayarConv2D, and SRMConv2D first-layer features (ablation) | IMC-7 · IMC-7 validation | accuracy: 92.1% -> 93.4% | Helped |
| Use sequential time-axis ConvLSTM analysis for multi-resolution Z-score maps (ablation) | LADN · Anomaly validation overall | F1: 41.93% -> 48.39% | Helped |
| Add multi-resolution Z-score windows to global Z-score features (ablation) | LADN · Anomaly validation overall | F1: 33.81% -> 41.93% | Helped |
| Replace Z-score normalization with raw feature differences (ablation) | LADN · Anomaly validation overall | F1: 33.81% -> 14.61% | Hurt |
| Initialize the manipulation-trace extractor from IMC-385 and allow low-rate updates (ablation) | ManTra-Net · Photoshop-battle dataset | image-level AUC: 61.85% -> 75.88% | Helped |
| Freeze the pretrained IMC-385 feature extractor (ablation) | ManTra-Net · Deep-inpainting dataset | F1: 68.61% -> 48.39% | Hurt |
Closest papers by meaning
- Reinforced Multi-teacher Knowledge Distillation for Efficient General Image Forgery Detection and Localization (AAAI 2025 · similarity 0.89)
- Learning Discriminative Noise Guidance for Image Forgery Detection and Localization (AAAI 2024 · similarity 0.88)
- M²RL-Net: Multi-View and Multi-Level Relation Learning Network for Weakly-Supervised Image Forgery Detection (AAAI 2025 · similarity 0.87)
- Towards Modern Image Manipulation Localization: A Large-Scale Dataset and Novel Methods (CVPR 2024 · similarity 0.86)
- FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Models (ICLR 2025 · similarity 0.86)
- UnionFormer: Unified-Learning Transformer with Multi-View Representation for Image Manipulation Detection and Localization (CVPR 2024 · similarity 0.86)
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.