Open AbdullahKhanML opened 3 years ago
I have had the exact same thing happen - the sample video works perfectly (for imshow and imwrite), but my own video produces nothing. Did you manage to fix this Abdullah, or find another solution? Maybe the model needs to be re-trained based on the zoom/angle/light conditions of our own videos.
I ran the code with the input video provided in the directory and it absolutely worked fine, but whenever I try to input my own video, it doesn't create bounding boxes around vehicles and nor count them and speed also doesn't work.
Furthermore, the fps is too low with the input video, but that is maybe I am running the project on RPi 4B I do have a compute neural stick (NCS2) I would like to deploy this model onto the stick once it starts counting vehicles and measuring speed.