Method Details
Details for method 'Instance Segmentation by Jointly Optimizing Spatial Embeddings and Clustering Bandwidth'
Method overview
| name | Instance Segmentation by Jointly Optimizing Spatial Embeddings and Clustering Bandwidth |
| challenge | instance-level semantic labeling |
| details | Fine only - ERFNet backbone |
| publication | Instance Segmentation by Jointly Optimizing Spatial Embeddings and Clustering Bandwidth Davy Neven, Bert De Brabandere, Marc Proesmans and Luc Van Gool CVPR 2019 http://openaccess.thecvf.com/content_CVPR_2019/papers/Neven_Instance_Segmentation_by_Jointly_Optimizing_Spatial_Embeddings_and_Clustering_Bandwidth_CVPR_2019_paper.html |
| project page / code | https://github.com/davyneven/SpatialEmbeddings |
| used Cityscapes data | fine annotations |
| used external data | |
| runtime | 0.1 s Nvidia 1080 ti |
| subsampling | no |
| submission date | October, 2018 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| AP | 27.6526 |
| AP50% | 50.8803 |
| AP100m | 37.8283 |
| AP50m | 37.3337 |
Class results
| Class | AP | AP50% | AP100m | AP50m |
|---|---|---|---|---|
| person | 34.5089 | 65.0544 | 49.8588 | 49.3945 |
| rider | 26.0957 | 58.7783 | 36.9226 | 36.9831 |
| car | 52.4051 | 75.3141 | 71.1283 | 72.4886 |
| truck | 21.6719 | 33.1038 | 26.2447 | 24.2227 |
| bus | 31.1974 | 45.1526 | 43.5917 | 43.522 |
| train | 16.3798 | 32.4482 | 22.6754 | 19.577 |
| motorcycle | 20.0665 | 48.4046 | 25.1995 | 25.9785 |
| bicycle | 18.8954 | 48.7862 | 27.0055 | 26.5031 |