Research map / Video / Action recognition
Extreme Low Resolution Activity Recognition With Multi-Siamese Embedding Learning
Read the paperCheck 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 multi-Siamese embedding learning to two-stream CNN with multiple LR transforms (ablation) | Two-stream CNN · 16x12 HMDB | accuracy: 35.34 ± 0.41 -> 37.70 ± 0.17 | Helped |
| Add multi-Siamese embedding learning to two-stream CNN with multiple LR transforms (ablation) | Two-stream CNN · 16x12 DogCentric | accuracy: 68.09 -> 69.43 | Helped |
| Use a two-stream RGB-and-flow model instead of a one-stream RGB model (ablation) | Baseline CNN · 16x12 HMDB | accuracy: 25.08 ± 0.40 -> 31.50 ± 0.30 | Helped |
| Use a two-stream RGB-and-flow model instead of a one-stream RGB model (ablation) | Baseline CNN · 16x12 DogCentric | accuracy: 53.05 -> 61.25 | Helped |
| Add multiple LR transforms without embedding learning (ablation) | One-stream CNN · 16x12 HMDB | accuracy: 25.08 ± 0.40 -> 25.17 ± 0.24 | No clear effect |
| Add multiple LR transforms without embedding learning (ablation) | One-stream CNN · 16x12 DogCentric | accuracy: 53.05 -> 57.61 | Helped |
| Add multi-Siamese embedding learning (ablation) | One-stream CNN with multiple LR transforms · 16x12 HMDB | accuracy: 25.17 ± 0.24 -> 26.21 ± 0.27 | Helped |
| Add multi-Siamese embedding learning (ablation) | One-stream CNN with multiple LR transforms · 16x12 DogCentric | accuracy: 57.61 -> 59.08 | Helped |
| Add multi-Siamese embedding learning (ablation) | Two-stream CNN · Hardcore Henry | F1-score: 0.871 -> 0.885 | Helped |
Closest papers by meaning
- Privacy-Preserving Human Activity Recognition from Extreme Low Resolution (AAAI 2017 · similarity 0.89)
- HARDVS: Revisiting Human Activity Recognition with Dynamic Vision Sensors (AAAI 2024 · similarity 0.84)
- Fast Temporal Activity Proposals for Efficient Detection of Human Actions in Untrimmed Videos (CVPR 2016 · similarity 0.84)
- Semi-Supervised Action Recognition With Temporal Contrastive Learning (CVPR 2021 · similarity 0.84)
- Rethinking Resolution in the Context of Efficient Video Recognition (NeurIPS 2022 · similarity 0.84)
- Learning Human Action Recognition Representations Without Real Humans (NeurIPS 2023 · 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.