Method Details
Details for method 'PIDNet-S'
Method overview
| name | PIDNet-S |
| challenge | pixel-level semantic labeling |
| details | |
| publication | PIDNet: A Real-time Semantic Segmentation Network Inspired from PID Controller https://arxiv.org/abs/2206.02066 |
| project page / code | https://github.com/XuJiacong/PIDNet |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 0.0107 s RTX 3090, Anaconda Env, I7-11700K |
| subsampling | no |
| submission date | May, 2022 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 78.6099 |
| iIoU Classes | 54.9503 |
| IoU Categories | 90.4565 |
| iIoU Categories | 77.9184 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.5267 | - |
| sidewalk | 85.5553 | - |
| building | 92.8102 | - |
| wall | 52.8046 | - |
| fence | 58.9667 | - |
| pole | 65.4784 | - |
| traffic light | 73.7669 | - |
| traffic sign | 76.4122 | - |
| vegetation | 93.4628 | - |
| terrain | 71.8358 | - |
| sky | 95.573 | - |
| person | 85.4232 | 67.2542 |
| rider | 70.3374 | 46.6057 |
| car | 95.5493 | 89.153 |
| truck | 68.1144 | 32.3341 |
| bus | 84.2476 | 47.0611 |
| train | 85.1703 | 53.0852 |
| motorcycle | 64.7053 | 43.2524 |
| bicycle | 74.8476 | 60.8568 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.6719 | - |
| nature | 93.1535 | - |
| object | 71.7628 | - |
| sky | 95.573 | - |
| construction | 93.2857 | - |
| human | 85.674 | 68.5063 |
| vehicle | 95.0744 | 87.3306 |
