Research map / Video / Action recognition
Generalized Rank Pooling for Activity Recognition
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 temporal ordering constraints to the low-rank subspace representation (ablation) | GRP · JHMDB (3 splits) | Avg. Acc. (%): 64.1 -> 70.6 | Helped |
| Add temporal ordering constraints to the low-rank subspace representation (ablation) | GRP · MPII Cooking Activities (7 splits) | mAP (%): 66.1 -> 68.4 | Helped |
| Add temporal ordering constraints to the low-rank subspace representation (ablation) | GRP · HMDB-51 (3 splits) | Avg. Acc. (%): 63.1 -> 65.4 | Helped |
| Add temporal ordering constraints to the low-rank subspace representation (ablation) | GRP · UCF101 (3 splits) | Avg. Acc. (%): 90.1 -> 91.9 | Helped |
| Estimate subspace basis vectors incrementally instead of jointly (ablation) | GRP · JHMDB (split-1) | Avg. Acc. (%): 73.8 -> 74.3 | Mixed |
| Use FC6 rather than FC7 features (ablation) | GRP · JHMDB (split-1) | Avg. Acc. (%): 72.0 -> 73.8 | Helped |
| Use FC6 rather than FC7 features (ablation) | GRP · MPII Cooking Activities (split-1) | mAP (%): 50.7 -> 53.8 | Helped |
| Use the RBF Projection Metric kernel rather than the linear kernel (ablation) | GRP classifier · MPII Cooking Activities (split-1) | mAP (%): 24.2 -> 52.1 | Helped |
| Use the RBF Projection Metric kernel rather than the linear kernel (ablation) | GRP classifier · JHMDB (split-1) | Avg. Acc. (%): 46.6 -> 66.8 | Helped |
| Use ResNet-152 rather than VGG features (ablation) | GRP + IDT-FV · HMDB-51 (3 splits) | Avg. Acc. (%): 67.0 -> 72.1 | Helped |
| Use ResNet-152 rather than VGG features (ablation) | GRP + IDT-FV · UCF101 (3 splits) | Avg. Acc. (%): 92.3 -> 93.5 | Helped |
| Use more subspaces in the GRP representation (ablation) | GRP · JHMDB (split-1) | accuracy | Mixed |
Closest papers by meaning
- Non-Linear Temporal Subspace Representations for Activity Recognition (CVPR 2018 · similarity 0.93)
- Discriminative Hierarchical Rank Pooling for Activity Recognition (CVPR 2016 · similarity 0.93)
- Learning End-to-end Video Classification with Rank-Pooling (ICML 2016 · similarity 0.90)
- Dynamic Image Networks for Action Recognition (CVPR 2016 · similarity 0.89)
- Video Representation Learning Using Discriminative Pooling (CVPR 2018 · similarity 0.88)
- AdaScan: Adaptive Scan Pooling in Deep Convolutional Neural Networks for Human Action Recognition in Videos (CVPR 2017 · 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.