You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
You may resize the image to keep the length of feature vector consistent before you extract the HOG features. e.g. add a line 'im = cv2.imresize(im, (150, 300))' after #38 and #48 respectively. Also, add 'import cv2' in the beginning.
I train my own data on windows,python3.6 , 192 pos images(RGB) and 192 neg images(RGB), size=(192,384)
[hog]
min_wdw_sz: [150, 300]
step_size: [8, 8]
orientations: 9
pixels_per_cell: [8, 8]
cells_per_block: [2, 2]
visualize: False
transform_sqrt:True
[nms]
threshold: 0.3
how to fix it?
The text was updated successfully, but these errors were encountered: