totosan / iotedgeobjecttracking Goto Github PK

For getting better results on smaller images to detect, what is neccessary, a specialized version is needed.
Maybe custom vision can help, otherwise retraining Yolo and converting it to ONNX can be a good way
is to be defined

If there are parking cars or other standing objectes, they should be ignored for a while.
It needs to check from time to time, if the object moves.

Not moving objects should not be marked in the image and not be send to the cloud as telemetry

with a good IP Cam (LUPS 202) I can collect images in the night. Detecting special objects could be a benefit, but needs object detection ML Model, that can do the job.
Yolo is not accurate in this scenario

for displaying object directions in the image, there should be a check box in the form for show or hide lines.
To indicate a direction, draw a line from center of object to next position