c++ - OpenCV trained cascade minneighbors set to 100 -


i train cascade classifier detect pedestrian traffic lights, used 5000 positive photos , 1000 negative training, here result xml file got.

the problem trained cascade detects many false images. used original code of opencv face detection, , notice best result when set minneighbors in detectmltiscale 100.

how should increase trained cascade quality?


Comments

Popular posts from this blog

ssh - Vagrant Windows - ssh_exchange_identification: read: Connection reset by peer -

gcc - Neither ld wrap nor LD_PRELOAD working to intercept system call -

I could not debug scala remote application on eclipse or IntelliJ IDEs -