Details for method 'DLA_HRNet48OCR_MSFLIP_000'
Method overview
| name |
DLA_HRNet48OCR_MSFLIP_000 |
| challenge |
pixel-level semantic labeling |
| details |
This set of predictions is from DLA (differentiable lattice assignment network) with "HRNet48+OCR-Head" as base segmentation model. The model is, first trained on coarse-data, and then trained on fine-annotated train/val sets. Multi-scale (0.5, 0.75, 1.0, 1.25, 1.5, 1.75) and flip scheme is adopted during inference. |
| publication |
Anonymous
|
| project page / code |
|
| used Cityscapes data |
fine annotations, coarse annotations |
| used external data |
|
| runtime |
n/a |
| subsampling |
no |
| submission date |
January, 2022 |
| previous submissions |
|
Average results
| Metric |
Value |
| IoU Classes | 84.8466 |
| iIoU Classes | 68.5869 |
| IoU Categories | 93.0071 |
| iIoU Categories | 84.4884 |
Class results
| Class |
IoU |
iIoU |
| road | 98.9388 | - |
| sidewalk | 89.176 | - |
| building | 94.7968 | - |
| wall | 71.0248 | - |
| fence | 67.5129 | - |
| pole | 75.5322 | - |
| traffic light | 81.9231 | - |
| traffic sign | 84.926 | - |
| vegetation | 94.3753 | - |
| terrain | 74.7301 | - |
| sky | 96.2152 | - |
| person | 89.7338 | 76.8513 |
| rider | 78.9152 | 62.0114 |
| car | 96.9069 | 92.34 |
| truck | 80.1325 | 55.4671 |
| bus | 93.0491 | 67.9426 |
| train | 87.0089 | 63.2497 |
| motorcycle | 76.2575 | 60.141 |
| bicycle | 80.93 | 70.6923 |
Category results
| Category |
IoU |
iIoU |
| flat | 98.8938 | - |
| nature | 94.1639 | - |
| object | 80.563 | - |
| sky | 96.2152 | - |
| construction | 94.8241 | - |
| human | 89.8247 | 77.815 |
| vehicle | 96.565 | 91.1617 |
Links
Download results as .csv file
Benchmark page
