how do I train SVM on images with different aspect ratios?

asked 2013-05-13 22:32:25 -0600

cb20130410
21 ●1 ●2

Can the SVM work on data with different dimensions? (using OpenCV's svm or libsvm) If the images have different size, I can resize to a standard value. But if they have different aspect ratios, it seems not to make sense to resize without keeping the original aspect ratio. Or shall I pad them with zeros to make them have the same aspect ratio?

edit retag flag offensive close merge delete

add a comment

1 answer

Sort by » oldest newest most voted

answered 2013-05-14 03:23:39 -0600

Guanta
6736 ●6 ●25 ●79

All training methods require to have the same feature sizes. Thus, if you use your images as features (maybe you should consider to change that), then they need to have the same size - zero-padding seems adequate.