Method Details
Details for method 'iFLYTEK-CV'
Method overview
| name | iFLYTEK-CV |
| challenge | pixel-level semantic labeling |
| details | Both fine(train&val) and coarse data were used to train a novel segmentation framework. |
| publication | IFLYTEK RESEARCH IFLYTEK CV Group - YinLin |
| project page / code | |
| used Cityscapes data | fine annotations, coarse annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | October, 2017 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 81.3516 |
| iIoU Classes | 60.8642 |
| IoU Categories | 91.033 |
| iIoU Categories | 79.4948 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.6123 | - |
| sidewalk | 86.3038 | - |
| building | 93.4406 | - |
| wall | 61.6008 | - |
| fence | 64.1701 | - |
| pole | 67.1773 | - |
| traffic light | 75.5251 | - |
| traffic sign | 79.207 | - |
| vegetation | 93.626 | - |
| terrain | 72.9594 | - |
| sky | 95.6405 | - |
| person | 86.3879 | 69.5608 |
| rider | 70.3213 | 49.2622 |
| car | 96.02 | 89.9997 |
| truck | 77.3383 | 46.5387 |
| bus | 90.9475 | 56.7789 |
| train | 88.342 | 58.8971 |
| motorcycle | 71.1026 | 50.866 |
| bicycle | 76.9574 | 65.0107 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.6734 | - |
| nature | 93.318 | - |
| object | 73.7221 | - |
| sky | 95.6405 | - |
| construction | 93.686 | - |
| human | 86.6406 | 70.5013 |
| vehicle | 95.5505 | 88.4883 |