Research map / Video / Video segmentation and tracking
Semantic Video CNNs Through Representation Warping
Raghudeep Gadde, Varun Jampani, Peter V. Gehler
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 the FlowCNN flow transformation to WarpNet modules at FC6+FC7 (ablation) | PlayData-CNN + WarpNet · CamVid test | IoU: 69.8 -> 70.2 | Improved |
| insert a WarpNet module at a single layer such as FC6 (ablation) | PlayData-CNN · CamVid test | IoU: 68.9 -> 69.8 | Improved |
| use two WarpNet modules (FC6 and FC7) instead of one at FC6 (ablation) | PlayData-CNN + WarpNet · CamVid test | IoU: 69.8 -> 70.2 | Improved |
| replace WarpNet with plain concatenation of FC6 and FC7 features across frames (ablation) | PlayData-CNN · CamVid test | IoU: 0.1% vs 1.3% | Hurt |
| use slower, more accurate FlowFields instead of DIS-Flow (ablation) | PlayData-CNN + WarpNet · CamVid test | IoU: 70.2 -> 70.3 | No clear effect |
| add WarpNet to Dilation-CNN at its 8 dilated convolutions | Dilation-CNN · CamVid test | IoU: 65.3 -> 67.1 | Improved |
| add WarpNet to single-scale PSPNet | PSPNet-SSc · Cityscapes val | IoU: 79.4 -> 80.6 | Improved |
| add WarpNet to multi-scale PSPNet | PSPNet-MSc · Cityscapes val | IoU: 80.8 -> 81.5 | Improved |
| add WarpNet to multi-scale PSPNet | PSPNet-MSc · Cityscapes val | tIoU: 41.5 -> 43.3 | Improved |
| add WarpNet to PSPNet-MSc | PSPNet-MSc · Cityscapes test, pole class IoU | IoU: 64.5 -> 68.6 | Improved |
| add WarpNet to PSPNet-MSc | PSPNet-MSc · Cityscapes test, truck class IoU | IoU: 73.6 -> 72.9 | Hurt |
| add WarpNet to PSPNet-MSc | PSPNet-MSc · Cityscapes test | iIoU: 58.1 -> 59.5 | Improved |
Closest papers by meaning
- Semantic Video Segmentation by Gated Recurrent Flow Propagation (CVPR 2018 · similarity 0.91)
- Temporally Distributed Networks for Fast Video Semantic Segmentation (CVPR 2020 · similarity 0.89)
- Low-Latency Video Semantic Segmentation (CVPR 2018 · similarity 0.88)
- Every Frame Counts: Joint Learning of Video Segmentation and Optical Flow (AAAI 2020 · similarity 0.88)
- Efficient Semantic Video Segmentation with Per-frame Inference (ECCV 2020 · similarity 0.88)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.87)
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.
