Method Details
Details for method 'SJTU_hpm'
Method overview
| name | SJTU_hpm |
| challenge | pixel-level semantic labeling |
| details | |
| publication | Hard Pixel Mining for Depth Privileged Semantic Segmentation Zhangxuan Gu, Li Niu*, Haohua Zhao, and Liqing Zhang |
| project page / code | |
| used Cityscapes data | fine annotations, coarse annotations, stereo |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | June, 2020 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 81.2524 |
| iIoU Classes | 59.0561 |
| IoU Categories | 91.4181 |
| iIoU Categories | 79.0908 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.7419 | - |
| sidewalk | 87.2696 | - |
| building | 93.6371 | - |
| wall | 61.6909 | - |
| fence | 61.872 | - |
| pole | 69.4222 | - |
| traffic light | 75.4598 | - |
| traffic sign | 80.7466 | - |
| vegetation | 93.7853 | - |
| terrain | 74.4005 | - |
| sky | 95.8965 | - |
| person | 86.7413 | 68.3369 |
| rider | 72.743 | 48.8074 |
| car | 96.222 | 90.8249 |
| truck | 74.9872 | 45.3645 |
| bus | 91.2507 | 55.5726 |
| train | 82.9468 | 49.4515 |
| motorcycle | 69.4609 | 51.3237 |
| bicycle | 76.5223 | 62.7673 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.7933 | - |
| nature | 93.5036 | - |
| object | 75.1602 | - |
| sky | 95.8965 | - |
| construction | 93.9208 | - |
| human | 86.8616 | 69.2102 |
| vehicle | 95.7908 | 88.9714 |