Research map / Video / Action recognition
SCSampler: Sampling Salient Clips From Video for Efficient Action Recognition
Bruno Korbar, Du Tran, Lorenzo Torresani
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 |
|---|---|---|---|
| select K=10 clips with SCSampler instead of dense evaluation | ir-CSN-152 · Sports1M | video accuracy (%): 77.0 -> 84.0 | Improved |
| select clips with SCSampler instead of uniform selection | MC3-18 · Sports1M | video accuracy (%): 64.8 -> 72.8 | Improved |
| select clips with SCSampler instead of dense evaluation | ir-CSN-152 · Sports1M | runtime (days): 14.0 -> 0.9 | Improved |
| select clips with SCSampler instead of dense evaluation | ir-CSN-152 · Kinetics | video accuracy (%): 78.8 -> 80.2 | Improved |
| use fully-convolutional spatial testing with SCSampler | ir-CSN-152 · Kinetics | video accuracy (%): 80.2 -> 81.1 | Improved |
| train visual sampler with AC loss instead of SAL-RANK (ablation) | Visual sampler · miniSports | video accuracy (%): 64.8 -> 73.1 | Improved |
| train audio sampler with SAL-RANK loss instead of AC (ablation) | Audio sampler · miniSports | video accuracy (%): 66.4 -> 67.8 | Improved |
| add I-frame features to MD+RGB-R (ablation) | Visual sampler ResNet-18 · miniSports | video accuracy (%): 73.1 -> 74.9 | Improved |
| use ResNet-18 instead of ShuffleNet-26 (ablation) | Visual sampler MD+RGB-R · miniSports | video accuracy (%): 67.9 -> 73.1 | Improved |
| combine MD and RGB-R features instead of MD alone (ablation) | Visual sampler ResNet-18 · miniSports | video accuracy (%): 63.5 -> 73.1 | Improved |
| add audio sampler via AV-union-list to visual sampler (ablation) | SCSampler · miniSports | video accuracy (%): 73.1 -> 76.0 | Improved |
| replace AV-union-list with joint training of both samplers (ablation) | SCSampler · miniSports | video accuracy (%): 76.0 -> 75.5 | No clear effect |
Later papers that compared with this method
8 later accepted papers report SCSampler as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- Text-Conditioned Resampler For Long Form Video Understanding (ECCV 2024 · better on 0 of 1 shared result)
- NSNet: Non-Saliency Suppression Sampler for Efficient Video Recognition (ECCV 2022 · better on 1 of 1 shared result)
- Temporal Saliency Query Network for Efficient Video Recognition (ECCV 2022 · better on 1 of 1 shared result)
- Dynamic Network Quantization for Efficient Video Inference (ICCV 2021 · better on 4 of 4 shared results)
- Efficient Action Recognition via Dynamic Knowledge Propagation (ICCV 2021 · better on 2 of 2 shared results)
- Selective Feature Compression for Efficient Activity Recognition Inference (ICCV 2021 · better on 0 of 1 shared result)
- Listen to Look: Action Recognition by Previewing Audio (CVPR 2020 · better on 2 of 2 shared results)
- AR-Net: Adaptive Frame Resolution for Efficient Action Recognition (ECCV 2020 · 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
- MGSampler: An Explainable Sampling Strategy for Video Action Recognition (ICCV 2021 · similarity 0.91)
- ActionAtlas: A VideoQA Benchmark for Domain-specialized Action Recognition (NeurIPS 2024 · similarity 0.89)
- Beyond Short Clips: End-to-End Video-Level Learning With Collaborative Memories (CVPR 2021 · similarity 0.89)
- No Frame Left Behind: Full Video Action Recognition (CVPR 2021 · similarity 0.89)
- Discriminative Video Representation Learning Using Support Vector Classifiers (TPAMI 2019 · similarity 0.89)
- Pulling Actions out of Context: Explicit Separation for Effective Combination (CVPR 2018 · similarity 0.88)
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.
