1 | initial version |
The open source world is moving massively to deep learning approaches without considering the "old school" methods are (still in 2020) the best trade-off in speed/accuracy for a lot of application (specially on ARM devices). I can suggest to use an earlier versions of OpenCV to train the cascades, I'm still using the traincascade in Opencv 2.4.10 and it is still perfectly working and suitable for OpenCV 3 and 4. More here:
Detecting people in cars it's tricky mostly due to lighting condition. Standard cascades in opencv won't work, you should try training custom cascades by yourself.
Regards!