Research map / Video / Action recognition
Knowledge Integration Networks for Action Recognition
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 the full two-stream KINet, with optical-flow TSN stream | Inception V3 action recognition · Kinetics-400 | top-1 accuracy: 76.6 -> 77.8 | Helped |
| Add KINet context integration | TSN-ResNet50 · Kinetics-400 | top-1 accuracy: 69.5 -> 72.4 (+2.9) | Helped |
| Add human parsing auxiliary task with knowledge distillation (ablation) | TSN-ResNet50 · Kinetics-400 | top-1 accuracy: 69.5 -> 70.3 (+0.8) | Helped |
| Add scene recognition auxiliary task with knowledge distillation (ablation) | TSN-ResNet50 · Kinetics-400 | top-1 accuracy: 69.5 -> 70.0 (+0.5) | Helped |
| Add human parsing and scene recognition auxiliary tasks (ablation) | TSN-ResNet50 · Kinetics-400 | top-1 accuracy: 69.5 -> 70.6 (+1.1) | Helped |
| Add CBI modules at res4 and res5 (ablation) | Multitask KINet · Kinetics-400 | top-1 accuracy: 70.6 -> 71.8 | Helped |
| Increase CBI count to two at res4 and one at res5 (ablation) | Multitask KINet · Kinetics-400 | top-1 accuracy: 71.8 -> 71.5 | Hurt |
| Use dot product instead of concatenation for AKG relations (ablation) | AKG KINet · Kinetics-400 | top-1 accuracy: 71.2 -> 71.7 | Helped |
| Use embedded dot product instead of concatenation for AKG relations (ablation) | AKG KINet · Kinetics-400 | top-1 accuracy: 71.2 -> 71.6 | Helped |
| Jointly learn auxiliary branch parameters instead of keeping them fixed (ablation) | KINet-ResNet50 · Kinetics-400 | top-1 accuracy: 70.5 -> 72.4 | Helped |
| Add an object branch through distillation (ablation) | KINet · Action recognition | performance | Hurt |
| Use a deeper GCN (ablation) | AKG KINet · Kinetics-400 | performance | No clear effect |
Closest papers by meaning
- I Know the Relationships: Zero-Shot Action Recognition via Two-Stream Graph Convolutional Networks and Knowledge Graphs (AAAI 2019 · similarity 0.88)
- VideoNet: A Large-Scale Dataset for Domain-Specific Action Recognition (CVPR 2026 · similarity 0.88)
- Convolutional Two-Stream Network Fusion for Video Action Recognition (CVPR 2016 · similarity 0.87)
- DVANet: Disentangling View and Action Features for Multi-View Action Recognition (AAAI 2024 · similarity 0.86)
- ActionAtlas: A VideoQA Benchmark for Domain-specialized Action Recognition (NeurIPS 2024 · similarity 0.86)
- Prompt-guided Disentangled Representation for Action Recognition (NeurIPS 2025 · similarity 0.85)
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.