Hi all
i am trying to create many samples using the perl script by Nataoshi to create 500 samples using positive images of quantity 6 and negative images of quantity 8. However the created output is a folder of 6 images . It seems not what i have expected which i thought i am supposed to get 500 images (or vector file)?
I've recently trained a Haar classifier by using 40 images, by following this here which is also inspired by Nataoshi.
The script I used to generate samples is:
perl bin/createsamples.pl positives.txt negatives.txt samples 1500 "opencv_createsamples -bgcolor 0 -bgthresh 0 -maxxangle 1.1 -maxyangle 1.1 maxzangle 0.5 -maxidev 40 -w 80 -h 40"
which results in having 1500 samples with the ratio of 80/40.
Although this is correct, making artificial data will not improve your model. Better to get 100 natural samples than 10000 artificial ones :)
Asked: 2017-07-22 13:17:26 -0600
Seen: 327 times
Last updated: Jul 26 '17
