Research map / Video / Video segmentation and tracking
Spatio-temporal Prompting Network for Robust Video Feature Extraction
Guanxiong Sun, Chi Wang, Zhaoyu Zhang, Jiankang Deng, Stefanos Zafeiriou, Yang Hua
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 transformer-based DVP predictor instead of Mixer-based (ablation) | STPN · ImageNet VID val (Swin-T) | AP50: 84.5 -> 85.0 | Improved |
| add STPN video prompts to FasterRCNN baseline | FasterRCNN Swin-T · ImageNet VID val | AP50: 80.7 -> 85.0 | Improved |
| add STPN to MixFormer tracker | MixFormer CVT-1K · GOT-10k | AO: 71.2 -> 72.7 | Improved |
| add STPN to MinVIS | MinVIS Swin-T · YouTube VIS 2019 | AP: 51.9 -> 54.8 | Improved |
| apply prompts at shallow layer only instead of deep (intermediate layers) (ablation) | STPN · ImageNet VID val (Swin variants) | AP | Improved |
| increase temporal stride S from 1 to 8 (ablation) | STPN · ImageNet VID val (Swin-T) | AP: 58.7 -> 60.6 | Improved |
| increase support frames K from 1 to 7 (ablation) | STPN · ImageNet VID val (Swin-T) | AP: 52.6 -> 60.6 | Improved |
| increase support frames K from 7 to 11 (ablation) | STPN · ImageNet VID val (Swin-T) | AP: 60.6 -> 60.7 (FPS 22.3 -> 20.5) | No clear effect |
| increase number of dynamic video prompts beyond 7 (ablation) | STPN · ImageNet VID val (Swin-T) | AP: 60.6 -> 60.6 (FPS 22.3 -> 19.9 at 13) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- MinVIS — MinVIS: A Minimal Video Instance Segmentation Framework without Video-based Training (NeurIPS 2022) · better on 1 of 2 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
- State Space Prompting via Gathering and Spreading Spatio-Temporal Information for Video Understanding (NeurIPS 2025 · similarity 0.88)
- STOP: Integrated Spatial-Temporal Dynamic Prompting for Video Understanding (CVPR 2025 · similarity 0.88)
- Hierarchical Visual Prompt Learning for Continual Video Instance Segmentation (ICCV 2025 · similarity 0.88)
- StNet: Local and Global Spatial-Temporal Modeling for Action Recognition (AAAI 2019 · similarity 0.86)
- Stable Mean Teacher for Semi-supervised Video Action Detection (AAAI 2025 · similarity 0.85)
- Learning Conditional Space-Time Prompt Distributions for Video Class-Incremental Learning (CVPR 2025 · similarity 0.85)
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.
