Research map / Video / Action recognition
Temporal Residual Networks for Dynamic Scene 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 |
|---|---|---|---|
| transform spatial ResNet into T-ResNet by adding temporal residual units | ResNet-50 · YUP++ | accuracy: 85.9 -> 89.0 | Helped |
| transform spatial ResNet into T-ResNet by adding temporal residual units | ResNet-50 · UCF101 | accuracy: 91.7% -> 93.9% | Helped |
| transform spatial ResNet into T-ResNet by adding temporal residual units | ResNet-50 · HMDB51 | accuracy: 61.2% -> 67.2% | Helped |
| add IDT-FV trajectory scores | T-ResNet two-stream · UCF101 | accuracy: 93.9% -> 94.6% | Helped |
| add IDT-FV trajectory scores | T-ResNet two-stream · HMDB51 | accuracy: 67.2% -> 70.6% | Helped |
| use temporal max pooling instead of temporal averaging (ablation) | T-ResNet · Dynamic scene classification | accuracy: ≈2% accuracy gain | Helped |
| start temporal max pooling earlier in the network (ablation) | T-ResNet · Dynamic scene classification | accuracy: ≈1−6% | Hurt |
| replace padded temporal convolutions with valid temporal convolutions (ablation) | T-ResNet · Dynamic scene classification | error: 2.4% | Hurt |
| evaluate videos with camera motion rather than static-camera videos | video recognition algorithms · YUP++ | accuracy | Hurt |
Closest papers by meaning
- Spatiotemporal Residual Networks for Video Action Recognition (NeurIPS 2016 · similarity 0.90)
- Temporal Convolutional Networks for Action Segmentation and Detection (CVPR 2017 · similarity 0.89)
- Attentive Temporal Pyramid Network for Dynamic Scene Classification (AAAI 2019 · similarity 0.89)
- T-C3D: Temporal Convolutional 3D Network for Real-Time Action Recognition (AAAI 2018 · similarity 0.88)
- Dynamic Filter Networks (NeurIPS 2016 · similarity 0.88)
- KernelWarehouse: Rethinking the Design of Dynamic Convolution (ICML 2024 · similarity 0.88)
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.