OpenCV 3v: problem with trained models

asked 2018-04-13 07:16:37 -0600

Hello all,

Situation : I use opencv 3, and train model with params

numStages 12 -minhitrate 0.999 -maxFalseAlarmRate 0.4 -numPos 50 -numNeg 320 -w 10 -h 10 -mode ALL -precalcBufSize 6000 -precalcIdxBufSize 6000 -featureType LBP

Results of trained model: it correctly recognizes object from camera, but with some issue: model recognizes object on image if I do small motions (maybe i create so blurring). What's problem it can be?

edit retag flag offensive close merge delete

add a comment

1 answer

Sort by » oldest newest most voted

answered 2018-04-18 02:12:18 -0600

StevenPuttemans

20029 ●16 ●82 ●207 http://stevenputtemans...

Your problem can be two-fold

It is the -w 10 -h 10 object size setting. This only gives you 100 pixels to generally describe your object class. It might be possible that small motion blurs simply remove to much content.
There was no blurring in your training data, and thus, your classifier overfitted to clean and clear data. To avoid that, you will need to add slightly blurred training samples to your training set.