Create positive images for rotated images for cascading training in OpenCV

I need to train a cascading classifier to detect vehicles and different viewing angles. I am using OpenCV.

Some of the angles I need to capture lead to the diagonal placement of the vehicle inside the image, as shown below:

enter image description here

, , , , , . / ? , . , . , , , , , , . , , - .

, ? , /, ?

enter image description here

+2
1

Q: , ?

A: . , , . ( - ). , 45 ° Haar- "", , .

, . , , , . .

:

A: . , .

( ), , OpenCV rotate(). 1 ° 360 ° (, 10 ° ) . , , .

, , . , .

, , , (, ) . , ( , , ).

, , , .

: http://docs.opencv.org/modules/objdetect/doc/latent_svm.html ( . , . )

http://www.araa.asn.au/acra/acra2006/papers/paper_5_63.pdf ( , . , , , .)

, . (: , - .

+6

Source: https://habr.com/ru/post/1523351/


All Articles