Research map / Detection and segmentation / CNN features
Object Detection in Video with Spatiotemporal Sampling Networks
Gedas Bertasius, Lorenzo Torresani, Jianbo Shi
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 temporal spatiotemporal sampling to the static baseline (ablation) | SSN · ImageNet VID val | mAP@0.5: 76.0 -> 78.9 | Improved |
| replace FGFA flow-based alignment with learned sampling | FGFA · ImageNet VID val | mAP@0.5: 78.8 -> 78.9 | Improved |
| use STSN instead of D&T without post-processing | D&T · ImageNet VID val | mAP@0.5: 75.8 -> 78.9 | Improved |
| add Seq-NMS temporal post-processing to STSN (ablation) | STSN · ImageNet VID val | mAP@0.5: 78.9 -> 80.4 | Improved |
| use STSN+ instead of D&T+ | D&T+ · ImageNet VID val | mAP@0.5: 79.8 -> 80.4 | Improved |
| increase temporal stride from 1 to 2 (ablation) | STSN · ImageNet VID val | mAP@0.5: 78.9 -> 79.0 | Improved |
| increase temporal stride from 1 to 4 (ablation) | STSN · ImageNet VID val | mAP@0.5: 78.9 -> 77.9 | Hurt |
| add more supporting frames (ablation) | STSN · ImageNet VID val | mAP@0.5 | Improved |
Closest papers by meaning
- R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016 · similarity 0.87)
- Detect to Track and Track to Detect (ICCV 2017 · similarity 0.87)
- Video Object Detection with an Aligned Spatial-Temporal Memory (ECCV 2018 · similarity 0.86)
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016 · similarity 0.86)
- Learning Where to Focus for Efficient Video Object Detection (ECCV 2020 · similarity 0.86)
- Spatio-Temporal Self-Organizing Map Deep Network for Dynamic Object Detection From Videos (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.
