Research map / Video / Video segmentation and tracking
Deep RNN Framework for Visual Sequential Applications
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 |
|---|---|---|---|
| Replace the 1-layer LSTM with 15-layer CBM | VGG-based action recognition · UCF-101 first test split | accuracy: 71.1 -> 79.8 | Helped |
| Replace 15-layer ConvLSTM with 15-layer CBM | VGG-based action recognition · HMDB-51 first test split | accuracy: 34.2 -> 40.2 | Helped |
| Use 15-layer CBM instead of the shallow LSTM | VGG-based action anticipation · UCF-101 first test split | accuracy: 30.6 -> 57.7 | Helped |
| Use 15-layer CBM instead of the shallow LSTM | VGG-based action anticipation · HMDB-51 first test split | accuracy: 18.8 -> 32.1 | Helped |
| Replace the shallow recurrent part with 15-layer CBM | UCF-101 action recognition with InceptionV1 · UCF-101 first test split | accuracy: 81.0 -> 85.3 | Helped |
| Replace the MCnet motion encoder's shallow ConvLSTM with 15-layer CBM | MCnet video prediction · KTH, average over 20 frames | PSNR: 25.9 -> 26.5 | Helped |
| Replace the MCnet motion encoder's shallow ConvLSTM with 15-layer CBM | MCnet video prediction · KTH, average over 20 frames | SSIM: 0.804 -> 0.824 | Helped |
| Use 15-layer CBM instead of 15-layer ConvLSTM | Polygon-RNN · Cityscapes validation | IoU (%): 46.7 -> 70.4 | Helped |
| Increase CBM depth from 5 to 15 layers (ablation) | Polygon-RNN · Cityscapes validation | IoU (%): 63.1 -> 70.4 | Helped |
| Replace element-wise production merge with addition (ablation) | 15-layer CBM · UCF-101 action recognition | accuracy: 79.8 -> 77.4 | Hurt |
| Replace element-wise production merge with addition (ablation) | 15-layer CBM · UCF-101 action anticipation | accuracy: 57.7 -> 56.7 | Hurt |
| Set temporal dropout rate to 0.5 rather than 0.0 (ablation) | 15-layer CBM with VGG19 · UCF-101 action recognition | accuracy: 75.2 -> 79.8 | Helped |
Closest papers by meaning
- Making Convolutional Networks Recurrent for Visual Sequence Learning (CVPR 2018 · similarity 0.90)
- Structural-RNN: Deep Learning on Spatio-Temporal Graphs (CVPR 2016 · similarity 0.88)
- Temporal Pyramid Recurrent Neural Network (AAAI 2020 · similarity 0.86)
- Towards flexible perception with visual memory (ICML 2025 · similarity 0.84)
- Recurrent neural network dynamical systems for biological vision (NeurIPS 2024 · similarity 0.84)
- LongVILA: Scaling Long-Context Visual Language Models for Long Videos (ICLR 2025 · 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.