Information regarding semantic segmentation in Autonomous Driving
Here few data labeling tools that helps to label the data for semantic segmentation.
Data Labeling Tools
There are few pixel wise annotated labeled datasets are avilable publicly.
Publicly available datasets for urban driving scenes (Autonomous Driving)
Number of Labelled Images | Classes | Multiple Cities | Environment | |
---|---|---|---|---|
KITTI | 200 | 34 | No | Daylight |
Cityscapes | 3478 | 34 | Yes | Daylight |
Mapillary | 20k | 66 | Yes | Daylight, rain, snow, fog, haze, dawn, dusk and night |
Apollo Scape | 147k | 36 | No | Daylight, snow, rain, foggy |
BDD100K | 8000 | 19 | Yes | Daylight, rain, snow, fog, haze, dawn, dusk and night |
For detailed information https://autonomous-driving.org/2018/07/15/semantic-segmentation-datasets-for-urban-driving-scenes/
Pre-trained semantic segmentation network on the Cityscape dataset