Research map / Video / Video-language understanding
ActivityForensics: A Comprehensive Benchmark for Localizing Manipulated Activity in Videos
Peijun Bao, Anwei Luo, Gang Pan, Alex C. Kot, Xudong Jiang
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 TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics intra-domain | avg AP: 70.67 -> 75.05 (+4.38) | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics intra-domain | avg AR: 74.31 -> 77.15 (+2.84) | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics intra-domain | AP@0.95: +9.78 | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics open-world (Vidu) | avg AP: 77.82 -> 83.64 (+5.82) | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics open-world (Vidu) | AP@0.95: +11.98 | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics cross-domain A->B | avg AP: 67.18 -> 69.63 (+2.45) | Improved |
| add TADiff to ActionFormer | TADiff / ActionFormer · ActivityForensics cross-domain B->A | avg AP: +3.75 | Improved |
| add noise injection alone (ablation) | TADiff / ActionFormer · ActivityForensics intra-domain | avg AP: 70.67 -> 70.38 | Hurt |
| add noise injection alone (ablation) | TADiff / ActionFormer · ActivityForensics open-world (Vidu) | avg AP: 77.82 -> 79.75 (+1.93) | Improved |
| add denoising alone (ablation) | TADiff / ActionFormer · ActivityForensics intra-domain | avg AP: 70.67 -> 73.52 | Improved |
| add denoising alone (ablation) | TADiff / ActionFormer · ActivityForensics open-world (Vidu) | avg AP: 77.82 -> 80.10 | Improved |
| combine noise injection with denoising (full TADiff) (ablation) | TADiff / ActionFormer · ActivityForensics intra-domain | avg AP: 73.52 -> 75.05 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ActionFormer — ActionFormer: Localizing Moments of Actions with Transformers (ECCV 2022) · 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
- Diagnosing Error in Temporal Action Detectors (ECCV 2018 · similarity 0.88)
- Explainable Forensics of Manipulated Segments in Untrimmed Long Videos (ICML 2026 · similarity 0.87)
- Adversarial Background-Aware Loss for Weakly-supervised Temporal Activity Localization (ECCV 2020 · similarity 0.87)
- Benchmarking the Robustness of Temporal Action Detection Models Against Temporal Corruptions (CVPR 2024 · similarity 0.87)
- Generalizing Deepfake Video Detection with Plug-and-Play: Video-Level Blending and Spatiotemporal Adapter Tuning (CVPR 2025 · similarity 0.86)
- Boundary Denoising for Video Activity Localization (ICLR 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.
