Research map / Video / Action recognition
Second-order Temporal Pooling for Action Recognition
Anoop Cherian, Stephen Jay Gould
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 |
|---|---|---|---|
| add SMAID stream to RGB+FLOW (ablation) | RGB+FLOW, VGG-16 · MPII Cooking | mAP: 38.1 -> 39.5 | Improved |
| add SMAID stream to RGB+FLOW (ablation) | RGB+FLOW, VGG-16 · JHMDB | accuracy: 55.9 -> 62.6 | Improved |
| add SMAID stream to RGB+FLOW (ablation) | RGB+FLOW, ResNet-152 · HMDB-51 (split-1) | accuracy: 62.1 -> 63.5 | Improved |
| add SMAID stream to RGB+FLOW (ablation) | RGB+FLOW, ResNet-152 · UCF101 (split-1) | accuracy: 89.5 -> 91.0 | Improved |
| add SMAID stream to RGB+FLOW (ablation) | RGB+FLOW, VGG-16 · UCF101 (split-1) | accuracy: 88.5 -> 88.8 | Improved |
| add SMAID stream to RGB (ablation) | RGB stream, VGG-16 · MPII Cooking | mAP: 33.9 -> 38.4 | Improved |
| add SMAID stream to RGB (ablation) | RGB stream, VGG-16 · JHMDB | accuracy: 51.5 -> 62.0 | Improved |
| add SMAID stream to RGB (ablation) | RGB stream, ResNet-152 · HMDB-51 (split-1) | accuracy: 45.4 -> 55.5 | Improved |
| add SMAID stream to RGB (ablation) | RGB stream, ResNet-152 · UCF101 (split-1) | accuracy: 83.1 -> 86.7 | Improved |
| replace SMAID stream with optical flow stream | RGB+X fusion, ResNet-152 · HMDB-51 (split-1) | accuracy: 59.5 -> 42.3 | Hurt |
| increase number of random permutations in BKCP from a single one to 32 (ablation) | BKCP on ResNet-152 features · TCP descriptor variance | average descriptor variance: 0.31 -> 1.5 | Hurt |
Closest papers by meaning
- Rank Pooling for Action Recognition (TPAMI 2016 · similarity 0.92)
- Video Representation Learning Using Discriminative Pooling (CVPR 2018 · similarity 0.92)
- Discriminatively Learned Hierarchical Rank Pooling Networks (IJCV 2017 · similarity 0.92)
- Discriminative Video Representation Learning Using Support Vector Classifiers (TPAMI 2019 · similarity 0.91)
- Learning End-to-end Video Classification with Rank-Pooling (ICML 2016 · similarity 0.91)
- AdaScan: Adaptive Scan Pooling in Deep Convolutional Neural Networks for Human Action Recognition in Videos (CVPR 2017 · similarity 0.90)
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.
