Research map / Video / Action recognition
Representing Sets of Instances for Visual Recognition
Jianxin Wu, Bin-Bin Gao, Guoqing Liu
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 |
|---|---|---|---|
| Replace VLAD with D3 on ITF instance vectors | VLAD · UCF 101 / HMDB 51 / Youtube | accuracy (%) | Improved |
| Replace VLAD (K=256) with D3 using half the dimensions (K=128) | VLAD · UCF 101 | accuracy (%): 82.54 -> 83.03 | Improved |
| Replace FV with D3 on ITF instance vectors | FV · UCF 101 / HMDB 51 | accuracy (%) | Hurt |
| Replace FV with D3 (K=256) | FV · Youtube | accuracy (%): 91.00 -> 91.55 | Improved |
| Add D3 to FV (hybrid D3+FV) (ablation) | FV · UCF 101, K=512 | accuracy (%): 85.23 -> 85.92 | Improved |
| Add D3 to FV (hybrid D3+FV) (ablation) | FV · HMDB 51, K=512 | accuracy (%): 58.13 -> 58.34 | Improved |
| Add D3 to FV (hybrid D3+FV) (ablation) | FV · Youtube, K=512 | accuracy (%): 91.00 -> 91.73 | Improved |
| Combine D3 rather than VLAD with FV (ablation) | VLAD+FV · UCF 101, K=512 | accuracy (%): 85.23 -> 85.92 | Improved |
| Use CNN (DSP) instance vectors: D3 instead of FV | FV · SUN 397, K=8 | accuracy (%): 57.97 -> 60.22 | Improved |
| Use CNN instance vectors: D3 instead of FV | FV · Caltech 101 / Caltech 256 | accuracy (%) | Hurt |
| Use CNN instance vectors: VLAD instead of D3 | D3 · SUN 397, K=8 | accuracy (%): 60.22 -> 61.22 | Improved |
| Use CNN instance vectors: D3+FV instead of single D3 (ablation) | D3 · Caltech 256, K=8 | accuracy (%): 82.92 -> 84.62 | Improved |
Closest papers by meaning
- Robust Regression (TPAMI 2016 · similarity 0.81)
- Transitive Invariance for Self-Supervised Visual Representation Learning (ICCV 2017 · similarity 0.81)
- DiVA: Diverse Visual Feature Aggregation for Deep Metric Learning (ECCV 2020 · similarity 0.81)
- Hierarchically Robust Representation Learning (CVPR 2020 · similarity 0.81)
- Submodular Attribute Selection for Visual Recognition (TPAMI 2016 · similarity 0.81)
- DFD: Distilling the Feature Disparity Differently for Detectors (ICML 2024 · similarity 0.80)
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.
