Method Details
Details for method 'GoogLeNet FCN'
Method overview
| name | GoogLeNet FCN |
| challenge | pixel-level semantic labeling |
| details | GoogLeNet No data augmentation, no graphical model Trained by Lukas Schneider, following "Fully Convolutional Networks for Semantic Segmentation", Long et al. CVPR 2015 |
| publication | Going Deeper with Convolutions Christian Szegedy , Wei Liu , Yangqing Jia , Pierre Sermanet , Scott Reed , Dragomir Anguelov , Dumitru Erhan , Vincent Vanhoucke , Andrew Rabinovich CVPR 2015 https://www.cs.unc.edu/~wliu/papers/GoogLeNet.pdf |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | January, 2017 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 63.002 |
| iIoU Classes | 38.6157 |
| IoU Categories | 85.7818 |
| iIoU Categories | 69.8078 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 97.4293 | - |
| sidewalk | 77.8929 | - |
| building | 89.1904 | - |
| wall | 35.0269 | - |
| fence | 38.975 | - |
| pole | 50.6238 | - |
| traffic light | 59.805 | - |
| traffic sign | 64.09 | - |
| vegetation | 91.2205 | - |
| terrain | 66.9271 | - |
| sky | 93.6679 | - |
| person | 76.1969 | 53.9986 |
| rider | 45.083 | 28.578 |
| car | 92.5667 | 85.0315 |
| truck | 33.3506 | 16.9454 |
| bus | 40.3751 | 29.5555 |
| train | 32.7402 | 19.2803 |
| motorcycle | 47.2575 | 25.7402 |
| bicycle | 64.6189 | 49.7958 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.1713 | - |
| nature | 90.902 | - |
| object | 58.6038 | - |
| sky | 93.6679 | - |
| construction | 89.4582 | - |
| human | 78.4386 | 56.3271 |
| vehicle | 91.231 | 83.2885 |