I have 2 questions:
1- I see there is segnet-camera.py (Cityscapes Segmentation) that only perform pixel segmentation. What if I want to get the detected model information like any other detection (bounding box, area, …)?
OR how to extract pixel info?
2- If it doesn’t give me the info that I want , how can I train Cityscapes to detect 21 models listed?
(PS: I am new to this so any elaboration is really appreciated)
1. Segmentation is a pixel-level detection.
If you want the information like bounding box, please check detectnet-camera.py.
2 A high level information can be found with detector directly.
If you want a high-level information from the pre-trained model, you can try to parse it
For example, the bounding box can be generated from max/min object pixel index.
Thanks for your response. Would you please Elaborate more how can I get some High level info like min/max object pixel index, or classID in python?