Details for method 'GridNet'
Method overview
| name |
GridNet |
| challenge |
pixel-level semantic labeling |
| details |
Conv-Deconv Grid-Network for semantic segmentation.
Using only the training set without extra coarse annotated data (only 2975 images).
No pre-training (ImageNet).
No post-processing (like CRF). |
| publication |
Anonymous
|
| project page / code |
|
| used Cityscapes data |
fine annotations |
| used external data |
|
| runtime |
n/a |
| subsampling |
no |
| submission date |
April, 2017 |
| previous submissions |
|
Average results
| Metric |
Value |
| IoU Classes | 69.4521 |
| iIoU Classes | 44.0627 |
| IoU Categories | 87.8573 |
| iIoU Categories | 71.1187 |
Class results
| Class |
IoU |
iIoU |
| road | 98.0331 | - |
| sidewalk | 82.7829 | - |
| building | 90.7833 | - |
| wall | 41.7756 | - |
| fence | 48.2926 | - |
| pole | 59.3055 | - |
| traffic light | 65.3798 | - |
| traffic sign | 69.4064 | - |
| vegetation | 92.3727 | - |
| terrain | 69.174 | - |
| sky | 93.7887 | - |
| person | 81.7599 | 57.2632 |
| rider | 62.2551 | 36.3841 |
| car | 93.0857 | 85.6167 |
| truck | 41.7916 | 21.4232 |
| bus | 56.2491 | 37.8033 |
| train | 49.0465 | 29.3036 |
| motorcycle | 55.2313 | 31.9642 |
| bicycle | 69.0767 | 52.7435 |
Category results
| Category |
IoU |
iIoU |
| flat | 98.402 | - |
| nature | 92.0528 | - |
| object | 65.4887 | - |
| sky | 93.7887 | - |
| construction | 90.9405 | - |
| human | 81.85 | 58.2589 |
| vehicle | 92.4782 | 83.9784 |
Links
Download results as .csv file
Benchmark page
