Research map / Video / Action recognition
A Key Volume Mining Deep Framework for Action 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 key-volume mining to the CNN using random spatial-temporal sampling | CNN action recognition · UCF101 split1 | accuracy (RGB stream): 82.1 -> 84.1 | Helped |
| Add key-volume mining to the CNN using random spatial-temporal sampling | CNN action recognition · UCF101 split1 | accuracy (flow stream): 85.4 -> 87.9 | Helped |
| Use unsupervised key-volume proposals instead of random spatial-temporal sampling (ablation) | Key-volume mining CNN, bag size 6 · UCF101 split1 | accuracy (RGB stream): 84.1 -> 84.8 | Helped |
| Use unsupervised key-volume proposals instead of random spatial-temporal sampling (ablation) | Key-volume mining CNN, bag size 6 · UCF101 split1 | accuracy (flow stream): 87.9 -> 89.0 | Helped |
| Increase bag size with unsupervised proposals from 1 to 6 (ablation) | Key-volume mining CNN · UCF101 split1 | accuracy (RGB stream): 82.2 -> 84.8 | Helped |
| Increase bag size with unsupervised proposals from 1 to 6 (ablation) | Key-volume mining CNN · UCF101 split1 | accuracy (flow stream): 86.4 -> 89.0 | Helped |
| Increase bag size from 6 to 12 with unsupervised proposals (ablation) | Key-volume mining CNN · UCF101 split1 | accuracy (RGB stream): 84.8 -> 85.3 | Helped |
| Increase bag size from 6 to 12 with unsupervised proposals (ablation) | Key-volume mining CNN · UCF101 split1 | accuracy (flow stream): 89.0 -> 89.2 | Helped |
| Replace stochastic out with max-out (ablation) | Key-volume mining CNN with unsupervised proposals, bag size 6 · UCF101 split1 | accuracy (RGB stream): 84.8 -> 84.5 | Hurt |
| Replace stochastic out with max-out (ablation) | Key-volume mining CNN with unsupervised proposals, bag size 6 · UCF101 split1 | accuracy (flow stream): 89.0 -> 88.5 | Hurt |
| Use random temporal sampling instead of random spatial-temporal sampling (ablation) | Key-volume mining CNN, bag size 6 · UCF101 split1 | accuracy (RGB stream): 84.1 -> 82.5 | Hurt |
| Use random temporal sampling instead of random spatial-temporal sampling (ablation) | Key-volume mining CNN, bag size 6 · UCF101 split1 | accuracy (flow stream): 87.9 -> 87.6 | Hurt |
Closest papers by meaning
- Mining 3D Key-Pose-Motifs for Action Recognition (CVPR 2016 · similarity 0.85)
- Unified Keypoint-Based Action Recognition Framework via Structured Keypoint Pooling (CVPR 2023 · similarity 0.85)
- Deep Sequential Context Networks for Action Prediction (CVPR 2017 · similarity 0.84)
- No Frame Left Behind: Full Video Action Recognition (CVPR 2021 · similarity 0.84)
- Unsupervised Deep Learning of Mid-Level Video Representation for Action Recognition (AAAI 2018 · similarity 0.84)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 2024 · similarity 0.81)
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.