Research map / Video / Video segmentation and tracking
STD2P: RGBD Semantic Segmentation Using Spatio-Temporal Data-Driven Pooling
Yang He, Wei-Chen Chiu, Margret Keuper, Mario Fritz
Check 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 superpixels to the single-frame FCN (ablation) | FCN · NYUDv2 40-class | Mean IoU: 34.0 -> 36.0 | Improved |
| Train with additional unlabeled frames (ablation) | STD2P superpixel · NYUDv2 40-class | Mean IoU: 36.0 -> 38.1 | Improved |
| Use multiple views at test time (ablation) | STD2P superpixel+ · NYUDv2 40-class | Mean IoU: 38.1 -> 40.1 | Improved |
| Replace average temporal pooling with max pooling (ablation) | STD2P full model · NYUDv2 40-class | Mean IoU: 40.1 -> 38.0 | Hurt |
| Replace average spatial pooling with max pooling (ablation) | STD2P full model · NYUDv2 40-class | Mean IoU: 40.1 -> 33.8 | Hurt |
| Replace both average pooling operations with max pooling (ablation) | STD2P full model · NYUDv2 40-class | Mean IoU: 40.1 -> 32.1 | Hurt |
| Use region correspondences rather than pixel correspondences (ablation) | Multi-view FCN segmentation · NYUDv2 40-class | Mean IoU: 34.6 -> 40.1 | Improved |
| Use wider and future views (ablation) | STD2P multi-view prediction · NYUDv2 40-class | Mean IoU | Improved |
| Add multi-view test-time information (ablation) | STD2P superpixel+ · SUN3D 33-class | Mean IoU: +3.5pp | Improved |
| Use superpixel-based partial-label training (ablation) | FCN · SUN3D 33-class | Mean IoU: +3.3pp | Improved |
| Use STD2P multi-view pooling (ablation) | STD2P full model · NYUDv2 4-class | Pixel Acc.: 82.7 -> 83.6 | Improved |
| Use STD2P multi-view pooling (ablation) | STD2P superpixel+ · NYUDv2 13-class | Mean Acc.: 67.0 -> 68.4 | Improved |
Closest papers by meaning
- 3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation (ECCV 2018 · similarity 0.89)
- Weakly Supervised Semantic Segmentation Using Superpixel Pooling Network (AAAI 2017 · similarity 0.88)
- Revisiting Superpixels for Active Learning in Semantic Segmentation With Realistic Annotation Costs (CVPR 2021 · similarity 0.88)
- Depth-aware CNN for RGB-D Segmentation (ECCV 2018 · similarity 0.87)
- Superpixel Convolutional Networks Using Bilateral Inceptions (ECCV 2016 · similarity 0.87)
- 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans (CVPR 2019 · 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.
