Research map / Detection and segmentation / CNN features
Fast Object Detection in Compressed Video
Shiyao Wang, Hongchao Lu, Zhidong Deng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 motion-vector warping to LSTM memory (ablation) | LSTM + residual · ImageNet VID val | mAP (%): 66.1 -> 72.1 | Improved |
| add motion-vector warping to LSTM only memory (ablation) | LSTM only · ImageNet VID val | mAP (%): 66.3 -> 71.3 | Improved |
| add residual errors without motion vectors (ablation) | LSTM memory · ImageNet VID val | mAP (%): 66.3 -> 66.1 | Hurt |
| add motion vectors to fast-moving objects (ablation) | LSTM memory · ImageNet VID val (fast motion) | mAP (%): 27.7 -> 38.5 | Improved |
| add residual errors on top of motion vectors and LSTM (ablation) | MV + LSTM · ImageNet VID val | mAP (%): 71.3 -> 72.1 | Improved |
| add LSTM to motion vectors and residual errors (ablation) | MV + residual · ImageNet VID val | mAP (%): 70.3 -> 72.1 | Improved |
| add LSTM to motion vectors and residual errors (ablation) | MV + residual · ImageNet VID val | speed (fps): 51.3 -> 41.7 | Hurt |
| add pyramidal feature attention (ablation) | MV + residual + LSTM · ImageNet VID val | mAP (%): 72.1 -> 73.0 | Improved |
| add pyramidal feature attention (ablation) | MV + residual + LSTM · ImageNet VID val (medium motion) | mAP (%): 72.0 -> 73.4 | Improved |
| add pyramidal feature attention (ablation) | MV + residual + LSTM · ImageNet VID val (slow motion) | mAP (%): 83.6 -> 84.7 | Improved |
| add pyramidal feature attention (ablation) | MV + residual + LSTM · ImageNet VID val (fast motion) | mAP (%): 44.2 -> 43.7 | Hurt |
| add pyramidal feature attention (ablation) | MV + residual + LSTM · ImageNet VID val | speed (fps): 41.7 -> 40.5 | Hurt |
Closest papers by meaning
- Object Guided External Memory Network for Video Object Detection (ICCV 2019 · similarity 0.87)
- Video Object Detection with an Aligned Spatial-Temporal Memory (ECCV 2018 · similarity 0.87)
- Object Detection From Video Tubelets With Convolutional Neural Networks (CVPR 2016 · similarity 0.87)
- Mobile Video Object Detection With Temporally-Aware Feature Maps (CVPR 2018 · similarity 0.87)
- Optimizing Video Object Detection via a Scale-Time Lattice (CVPR 2018 · similarity 0.86)
- Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors (CVPR 2017 · similarity 0.86)
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.
