Method Details
Details for method 'ESANet RGB (small input)'
Method overview
| name | ESANet RGB (small input) |
| challenge | pixel-level semantic labeling |
| details | ESANet: Efficient RGB-D Semantic Segmentation for Indoor Scene Analysis. ESANet-R34-NBt1D using RGB images with half the input resolution. |
| publication | Efficient RGB-D Semantic Segmentation for Indoor Scene Analysis Daniel Seichter, Mona Köhler, Benjamin Lewandowski, Tim Wengefeld and Horst-Michael Gross |
| project page / code | https://github.com/TUI-NICR/ESANet |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 0.031 s NVIDIA Jetson AGX Xavier (Jetpack 4.4, TensorRT 7.1, Float16) |
| subsampling | 2 |
| submission date | October, 2020 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 72.8739 |
| iIoU Classes | 40.5152 |
| IoU Categories | 87.0524 |
| iIoU Categories | 66.5409 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.2446 | - |
| sidewalk | 84.0601 | - |
| building | 91.1737 | - |
| wall | 57.1172 | - |
| fence | 52.56 | - |
| pole | 55.7002 | - |
| traffic light | 61.295 | - |
| traffic sign | 66.8496 | - |
| vegetation | 91.5616 | - |
| terrain | 69.6212 | - |
| sky | 94.5555 | - |
| person | 79.2713 | 49.6399 |
| rider | 62.754 | 30.9949 |
| car | 93.8732 | 85.6427 |
| truck | 64.9196 | 23.27 |
| bus | 71.6032 | 33.305 |
| train | 64.8048 | 32.9657 |
| motorcycle | 56.9792 | 25.7532 |
| bicycle | 67.6592 | 42.55 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.2522 | - |
| nature | 91.2202 | - |
| object | 61.5863 | - |
| sky | 94.5555 | - |
| construction | 91.3102 | - |
| human | 79.2822 | 50.5049 |
| vehicle | 93.16 | 82.5768 |