Research map / Detection and segmentation / CNN features
ShapeConv: Shape-Aware Convolutional Layer for Indoor RGB-D Semantic Segmentation
Jinming Cao, Hanchao Leng, Dani Lischinski, Daniel Cohen-Or, Changhe Tu, Yangyan Li
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 W_B only (no W_S) to the vanilla convolution (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 45.9 -> 47.0 | Improved |
| add W_S only (no W_B) to the vanilla convolution (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 45.9 -> 46.3 | Improved |
| add both W_B and W_S (full ShapeConv) (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 45.9 -> 47.4 | Improved |
| use HHA instead of raw depth as input (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 44.9 -> 45.9 | Improved |
| add depth input to RGB (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 43.9 -> 44.9 | Improved |
| add multi-scale testing (ablation) | Deeplabv3+ ResNet-101 (HHA) · NYUDv2-40 | Mean IoU: 45.9 -> 47.6 | Improved |
| replace vanilla conv by ShapeConv with Depth input (ablation) | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 44.9 -> 46.2 | Improved |
| replace vanilla conv by ShapeConv with HHA input | Deeplabv3+ ResNet-101 · NYUDv2-40 | Mean IoU: 45.9 -> 47.4 | Improved |
| replace vanilla conv by ShapeConv | Deeplabv3+ ResNet-50 · NYUDv2-13 | Mean IoU: 60.8 -> 61.8 | Improved |
| replace vanilla conv by ShapeConv | Deeplabv3+ ResNet-50 · SUN-RGBD | Mean IoU: 45.5 -> 46.3 | Improved |
| replace vanilla conv by ShapeConv | Deeplabv3+ ResNet-101 · SID | Mean IoU: 54.6 -> 60.6 | Improved |
| replace vanilla conv by ShapeConv | UNet ResNet-50 · NYUDv2-40 | Mean IoU: 39.7 -> 42.0 | Improved |
Closest papers by meaning
- Depth-aware CNN for RGB-D Segmentation (ECCV 2018 · similarity 0.90)
- Locality-Sensitive Deconvolution Networks With Gated Fusion for RGB-D Indoor Semantic Segmentation (CVPR 2017 · similarity 0.90)
- 3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation (ECCV 2018 · similarity 0.89)
- RDFNet: RGB-D Multi-Level Residual Feature Fusion for Indoor Semantic Segmentation (ICCV 2017 · similarity 0.89)
- DFormer: Rethinking RGBD Representation Learning for Semantic Segmentation (ICLR 2024 · similarity 0.88)
- DFormerv2: Geometry Self-Attention for RGBD Semantic Segmentation (CVPR 2025 · similarity 0.88)
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.
