Research map / Video / Action recognition
A Multi-Stream Bi-Directional Recurrent Neural Network for Fine-Grained Action Detection
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 |
|---|---|---|---|
| Use the four-stream CNN with pixel trajectories and a bi-directional LSTM instead of the strongest reported baseline | Fine-grained action detection · MPII Cooking 2 | mAP: 34.5% -> 41.2% | Helped |
| Use a multi-stream network with bi-directional LSTM instead of a two-stream network | Fine-grained action detection · MERL Shopping | mAP: 65.21% -> 80.31% | Helped |
| Replace stacked optical flow with pixel trajectories (ablation) | Person-centric motion stream · MPII Cooking 2 | mAP: 21.31% -> 22.35%; 27.36% -> 29.51% | Helped |
| Replace stacked optical flow with pixel trajectories (ablation) | Person-centric motion stream · MERL Shopping | mAP: 55.29% -> 57.88%; 71.70% -> 73.06% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric stacked-flow motion stream · MPII Cooking 2 | mAP: 21.31% -> 27.36% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric stacked-flow motion stream · MERL Shopping | mAP: 55.29% -> 71.70% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric pixel-trajectory motion stream · MPII Cooking 2 | mAP: 22.35% -> 29.51% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric pixel-trajectory motion stream · MERL Shopping | mAP: 57.88% -> 73.06% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric appearance stream · MPII Cooking 2 | mAP: 24.72% -> 28.77% | Helped |
| Add a unidirectional LSTM (ablation) | Person-centric appearance stream · MERL Shopping | mAP: 40.02% -> 63.26% | Helped |
| Add person-centric streams to the full-frame two-stream network (ablation) | Multi-stream action detector · MPII Cooking 2 | mAP: 30.18% -> 33.38% | Helped |
| Add person-centric streams to the full-frame two-stream network (ablation) | Multi-stream action detector · MERL Shopping | mAP: 65.21% -> 69.08% | Helped |
Closest papers by meaning
- MARS: Motion-Augmented RGB Stream for Action Recognition (CVPR 2019 · similarity 0.89)
- Spatiotemporal Residual Networks for Video Action Recognition (NeurIPS 2016 · similarity 0.88)
- Convolutional Two-Stream Network Fusion for Video Action Recognition (CVPR 2016 · similarity 0.88)
- Dance With Flow: Two-In-One Stream Action Detection (CVPR 2019 · similarity 0.87)
- Spatiotemporal Pyramid Network for Video Action Recognition (CVPR 2017 · similarity 0.87)
- Progressively Parsing Interactional Objects for Fine Grained Action Detection (CVPR 2016 · 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.