Method Details
Details for method 'DPN'
Method overview
| name | DPN |
| challenge | pixel-level semantic labeling |
| details | DPN trained on full resolution images |
| publication | Semantic Image Segmentation via Deep Parsing Network Z. Liu, X. Li, P. Luo, C. C. Loy, and X. Tang ICCV 2015 http://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Liu_Semantic_Image_Segmentation_ICCV_2015_paper.pdf |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | pre-training using ImageNet data |
| runtime | n/a |
| subsampling | no |
| submission date | April, 2016 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 66.8111 |
| iIoU Classes | 39.0655 |
| IoU Categories | 86.015 |
| iIoU Categories | 69.0931 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 97.476 | - |
| sidewalk | 78.4506 | - |
| building | 89.4665 | - |
| wall | 40.445 | - |
| fence | 45.8682 | - |
| pole | 51.1034 | - |
| traffic light | 56.8448 | - |
| traffic sign | 65.3499 | - |
| vegetation | 91.5435 | - |
| terrain | 69.4395 | - |
| sky | 94.4836 | - |
| person | 77.5429 | 53.5931 |
| rider | 54.2392 | 28.8877 |
| car | 92.4581 | 85.0371 |
| truck | 44.4937 | 20.1474 |
| bus | 53.4021 | 28.3354 |
| train | 49.9237 | 24.8667 |
| motorcycle | 52.0894 | 24.7784 |
| bicycle | 64.7905 | 46.8782 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.2115 | - |
| nature | 91.1392 | - |
| object | 58.9141 | - |
| sky | 94.4836 | - |
| construction | 89.7564 | - |
| human | 78.3879 | 55.047 |
| vehicle | 91.212 | 83.1392 |