Research map / Video / Video segmentation and tracking
Temporal Perceiver: A General Architecture for Arbitrary Boundary Detection
Jing Tan, Yuhong Wang, Gangshan Wu, Limin Wang
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 |
|---|---|---|---|
| use Temporal Perceiver instead of previous best method on shot boundary detection | Temporal Perceiver · SoccerNet-v2 | average-mAP: outperforms previous state of the art by 3.4% | Improved |
| use Temporal Perceiver on fading transitions | Temporal Perceiver · SoccerNet-v2 fading transitions | mAP: large margin of 25% | Improved |
| use Temporal Perceiver instead of previous SOTA on scene boundaries | Temporal Perceiver · MovieScenes | AP: improves the AP by 4.8% | Improved |
| use Temporal Perceiver instead of previous SOTA on scene boundaries | Temporal Perceiver · MovieScenes | Miou: improves Miou by 4.3% | Improved |
| use latent-query compression (M=0.5N) instead of zero compression (M=N) (ablation) | Temporal Perceiver · MovieNet | AP: improves the performance by 1.34% in AP and reduces the FLOPs by half | Improved |
| add alignment loss (ablation) | Temporal Perceiver · Kinetics-GEBD | f1@0.05: increases by 1.1% | Improved |
| add alignment loss (ablation) | Temporal Perceiver · Kinetics-GEBD | average F1: increases by 2.7% | Improved |
| add alignment loss (ablation) | Temporal Perceiver · MovieNet | Miou: increases by 1.3% | Improved |
| add alignment loss (ablation) | Temporal Perceiver · MovieNet | AP: increases by 2.7% | Improved |
| use learned boundary queries aligned to boundary features instead of directly concatenating boundary features with context centers (ablation) | Temporal Perceiver · Kinetics-GEBD and MovieNet | detection performance | Improved |
| use learned boundary likeliness at both training and inference instead of Gaussian kernels only at training (ablation) | Temporal Perceiver · coherence score ablation | detection performance | Improved |
| change input window length N by 2x or 3x larger (ablation) | Temporal Perceiver · default N per benchmark | detection performance | Hurt |
Later papers that compared with this method
3 later accepted papers report Temporal Perceiver as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Generic Event Boundary Detection via Denoising Diffusion (ICCV 2025 · better on 4 of 4 shared results)
- Neighbor Relations Matter in Video Scene Detection (CVPR 2024 · better on 1 of 1 shared result)
- Fine-grained Dynamic Network for Generic Event Boundary Detection (ECCV 2024 · better on 1 of 1 shared result)
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
- Boundary Flow: A Siamese Network That Predicts Boundary Motion Without Training on Motion (CVPR 2018 · similarity 0.88)
- Progressive Attention on Multi-Level Dense Difference Maps for Generic Event Boundary Detection (CVPR 2022 · similarity 0.88)
- UBoCo: Unsupervised Boundary Contrastive Learning for Generic Event Boundary Detection (CVPR 2022 · similarity 0.88)
- Local Compressed Video Stream Learning for Generic Event Boundary Detection (IJCV 2023 · similarity 0.87)
- Re-thinking Temporal Search for Long-Form Video Understanding (CVPR 2025 · similarity 0.87)
- Perception Encoder: The best visual embeddings are not at the output of the network (NeurIPS 2025 · 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.
