Research map / Video / Video segmentation and tracking
Switchable Temporal Propagation Network
Sifei Liu, Guangyu Zhong, Shalini De Mello, Jinwei Gu, Varun Jampani, Ming-Hsuan Yang, Jan Kautz
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 switchable TPN instead of basic TPN in fine-tuning on ACT (ablation) | TPN color propagation · ACT color (K=30) | RMSE: 6.04 -> 5.58 | Improved |
| use switchable TPN also in MS-COCO pretraining (ablation) | TPN color propagation · ACT color (K=30) | RMSE: 5.58 -> 5.55 | Improved |
| use switchable TPN in both stages vs basic TPN (ablation) | TPN color propagation · ACT color (K=10) | PSNR: 36.65 -> 37.64 | Improved |
| use switchable TPN in both stages vs basic TPN (ablation) | TPN color propagation · ACT color (K=40) | RMSE: 6.44 -> 5.99 | Improved |
| switchable TPN pretraining on top of switchable fine-tuning, PSNR (ablation) | TPN color propagation · ACT color (K=30) | PSNR: 35.26 -> 35.29 | No clear effect |
| reduce key-frame interval K from 40 to 10 | STPNim+STPNvd · ACT color | RMSE: 5.99 -> 3.98 | Improved |
| replace VPN with switchable TPN | color propagation · ACT color (K=30) | RMSE: 6.86 -> 5.55 | Improved |
| offline nearest-key-frame selection instead of online propagation | STPNim+STPNvd · ACT color (K=30) | RMSE: 5.55 -> 4.27 | Improved |
| use switchable TPN in both stages vs basic TPN, offline (ablation) | TPN color propagation · ACT color offline (K=30) | RMSE: 4.77 -> 4.27 | Improved |
| add SPN refinement to basic TPN segmentation (ablation) | SEG(t) · DAVIS | J-mean: 71.1 -> 76.19 | Improved |
| add SPN refinement to basic TPN segmentation (ablation) | SEG(t) · DAVIS | F-mean: 75.65 -> 73.53 | Hurt |
| replace VPN with TPN SEG(t) | segmentation propagation · DAVIS | J-mean: 70.2 -> 71.1 | Improved |
Closest papers by meaning
- Video Propagation Networks (CVPR 2017 · similarity 0.90)
- Memory Selection Network for Video Propagation (ECCV 2020 · similarity 0.86)
- Snap Video: Scaled Spatiotemporal Transformers for Text-to-Video Synthesis (CVPR 2024 · similarity 0.84)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.84)
- Blur-aware Spatio-temporal Sparse Transformer for Video Deblurring (CVPR 2024 · similarity 0.84)
- Direction-Aware Video Demoiréing with Temporal-Guided Bilateral Learning (AAAI 2024 · 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.
