Research map / Video / Action recognition
Active Vision for Early Recognition of Human Actions
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 mfIndRNN missing-frame estimation instead of a standard IndRNN (ablation) | IndRNN · NTU RGB-D, Test Scenario 1, View 1 | accuracy: 63.84 -> 70.53 | Helped |
| Add random frame-dropping data augmentation (ablation) | IndRNN · NTU RGB-D, Test Scenario 1, View 1 | accuracy: 63.84 -> 65.62 | Helped |
| Fill missing frames by linear interpolation (ablation) | IndRNN · NTU RGB-D, Test Scenario 1, View 1 | accuracy: 63.84 -> 66.03 | Helped |
| Add random frame-dropping data augmentation (ablation) | IndRNN · NTU RGB-D, Test Scenario 2, View 1 | accuracy: 73.27 -> 68.72 | Hurt |
| Fill missing frames by linear interpolation (ablation) | IndRNN · NTU RGB-D, Test Scenario 2, View 1 | accuracy: 73.27 -> 68.86 | Hurt |
| Use learned elapsed-time-dependent weights instead of uniform averaging (ablation) | Multiple-camera output integration · NTU RGB-D, Test Scenario 1 | accuracy: 75.35 -> 76.66 | Helped |
| Use learned elapsed-time-dependent weights instead of uniform averaging (ablation) | Multiple-camera output integration · NTU RGB-D, Test Scenario 2 | accuracy: 79.76 -> 81.28 | Helped |
| Learn a camera-selection policy from recurrent belief states rather than selecting views randomly | Multi-camera early action recognition · NTU RGB-D | acc: 54.81 -> 58.01 | Helped |
| Learn a camera-selection policy from recurrent belief states rather than selecting views randomly | Multi-camera early action recognition · IXMAS | acc: 92.21 -> 94.32 | Helped |
| Learn a modality-selection policy rather than selecting modalities randomly | Multi-modality early gesture recognition · nvGesture | acc: 62.29 -> 65.37 | Helped |
| Learn camera selection to avoid the severely occluded View 1 | Multi-camera early action recognition · NTU RGB-D, severe occlusion of View 1 | acc: 47.17 -> 53.30 | Helped |
Closest papers by meaning
- Deep Reinforcement Learning for Active Human Pose Estimation (AAAI 2020 · similarity 0.87)
- End-To-End Learning of Action Detection From Frame Glimpses in Videos (CVPR 2016 · similarity 0.86)
- Anticipating Visual Representations From Unlabeled Video (CVPR 2016 · similarity 0.86)
- Memory-Augmented Temporal Dynamic Learning for Action Recognition (AAAI 2019 · similarity 0.85)
- Evidential Active Recognition: Intelligent and Prudent Open-World Embodied Perception (CVPR 2024 · similarity 0.84)
- CanViT: Toward Active-Vision Foundation Models (NeurIPS 2026 · similarity 0.84)
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.