Research map / Video / Action recognition
A Deeper Dive Into What Deep Spatiotemporal Networks Encode: Quantifying Static vs. Dynamic Information
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 |
|---|---|---|---|
| add MATNet motion-to-appearance cross-connections | VOS fusion models · DAVIS16 analysis; MoCA evaluation | dynamic units | Helped |
| add the Fast pathway to SlowOnly | SlowFast action recognition · Kinetics-400 | dynamic units in the slow pathway: +3.3% | Helped |
| train FastOnly jointly with the Slow branch | SlowFast action recognition · Kinetics-400 | dynamic information extracted by the Fast branch | Helped |
| train on Something-Something-v2 rather than Diving48 | action recognition models · Something-Something-v2 and Diving48 | dynamic units | Helped |
| train on TAO-VOS rather than DAVIS16 or ImageNetVID | FusionSeg · VOS dataset-bias analysis | specialized dynamic units | Helped |
| remove boundary-aware refinement modules and boundary loss (ablation) | MATNet Ours · MoCA | mIoU: 67.3 -> 65.1 | Hurt |
| remove additional YouTube-VOS training (ablation) | MATNet Ours · MoCA | mIoU: 67.3 -> 54.7 | Hurt |
| remove dynamic-biased channels (ablation) | SlowFast trained on SSv2 · SSv2 validation | Top-1 accuracy | Hurt |
| remove the most bias-relevant channels rather than random channels (ablation) | MATNet and SlowFast · MoCA and SSv2 | task performance | Hurt |
Closest papers by meaning
- What Have We Learned From Deep Representations for Action Recognition? (CVPR 2018 · similarity 0.90)
- Spatiotemporal Pyramid Network for Video Action Recognition (CVPR 2017 · similarity 0.89)
- Spatiotemporal Fusion in 3D CNNs: A Probabilistic View (CVPR 2020 · similarity 0.89)
- Spatiotemporal Residual Networks for Video Action Recognition (NeurIPS 2016 · similarity 0.88)
- A Closer Look at Spatiotemporal Convolutions for Action Recognition (CVPR 2018 · similarity 0.88)
- A Large-Scale Study on Unsupervised Spatiotemporal Representation Learning (CVPR 2021 · similarity 0.87)
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.