Method Details
Details for method 'LDN-161'
Method overview
| name | LDN-161 |
| challenge | pixel-level semantic labeling |
| details | Ladder DenseNet-161 trained on train+val, fine labels only. Inference on multi-scale inputs. |
| publication | Efficient Ladder-style DenseNets for Semantic Segmentation of Large Images Ivan Kreso, Josip Krapac, Sinisa Segvic |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 2 s Titan Xp |
| subsampling | no |
| submission date | March, 2019 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 80.5992 |
| iIoU Classes | 56.4058 |
| IoU Categories | 91.2556 |
| iIoU Categories | 79.1492 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.6764 | - |
| sidewalk | 86.5149 | - |
| building | 93.5696 | - |
| wall | 61.8064 | - |
| fence | 60.9124 | - |
| pole | 68.2921 | - |
| traffic light | 75.5513 | - |
| traffic sign | 80.101 | - |
| vegetation | 93.7126 | - |
| terrain | 72.4412 | - |
| sky | 95.819 | - |
| person | 86.8302 | 68.1522 |
| rider | 72.2111 | 48.4005 |
| car | 96.1006 | 91.2092 |
| truck | 72.3375 | 39.097 |
| bus | 88.7623 | 50.1747 |
| train | 80.7035 | 48.8461 |
| motorcycle | 69.9344 | 43.8063 |
| bicycle | 77.1087 | 61.5604 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.7283 | - |
| nature | 93.3979 | - |
| object | 74.3379 | - |
| sky | 95.819 | - |
| construction | 93.7974 | - |
| human | 86.9561 | 69.1311 |
| vehicle | 95.7524 | 89.1673 |