Method Details
Details for method 'Roadstar.ai_CV(SFNet)'
Method overview
| name | Roadstar.ai_CV(SFNet) |
| challenge | pixel-level semantic labeling |
| details | same foucs net(SFNet), based on only fine labels, with focus on the loss distribution and same focus on the every layer of feature map |
| publication | Roadstar.ai-CV Maosheng Ye, Guang Zhou, Tongyi Cao, YongTao Huang, Yinzi Chen |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | |
| runtime | 0.2 s GTX1080ti |
| subsampling | no |
| submission date | December, 2017 |
| previous submissions | 1 |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 79.2054 |
| iIoU Classes | 60.823 |
| IoU Categories | 90.9881 |
| iIoU Categories | 82.5685 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.4046 | - |
| sidewalk | 85.3947 | - |
| building | 93.0442 | - |
| wall | 59.6 | - |
| fence | 59.2402 | - |
| pole | 67.5384 | - |
| traffic light | 76.3983 | - |
| traffic sign | 79.309 | - |
| vegetation | 93.7217 | - |
| terrain | 73.5724 | - |
| sky | 95.2812 | - |
| person | 86.7582 | 75.7695 |
| rider | 73.78 | 52.1544 |
| car | 95.7449 | 89.8975 |
| truck | 67.5296 | 48.8564 |
| bus | 81.2375 | 59.5406 |
| train | 72.118 | 44.8299 |
| motorcycle | 69.1618 | 47.771 |
| bicycle | 77.067 | 67.7646 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.6935 | - |
| nature | 93.4438 | - |
| object | 73.7795 | - |
| sky | 95.2812 | - |
| construction | 93.4814 | - |
| human | 86.963 | 76.3937 |
| vehicle | 95.2744 | 88.7433 |