ShapeConv: Shape-aware Convolutional Layer for Indoor RGB-D Semantic Segmentation

Benchmark Model Rank Results
semantic-segmentation-on-gamusShapeConv#4mIoU: 55.86
semantic-segmentation-on-llrgbd-syntheticShapeConv (ResNeXt-101)#6mIoU: 63.26
semantic-segmentation-on-nyu-depth-v2ShapeConv (ResNext-101)#39Mean IoU: 51.3%
semantic-segmentation-on-nyu-depth-v2ShapeConv (ResNet-101)#53Mean IoU: 49.0%
semantic-segmentation-on-nyu-depth-v2ShapeConv (ResNet-50)#55Mean IoU: 48.8%
semantic-segmentation-on-stanford2d3d-rgbdShapeConv-101#3mIoU: 60.6Pixel Accuracy: 82.7mAcc: 70.0
semantic-segmentation-on-sun-rgbdPSD-ResNet50#23Mean IoU: 48.6%
thermal-image-segmentation-on-rgb-t-glassShapeConv#10MAE: 0.054