Can I use lbp histogram for action recognition?
I am using openCV 3.0 with visual studio 2013. I have been trying to use SVM with LBP histogram, with 2 classes initially, but the predict value always shows same value for every input image. And I read somewhere that LBP can only be used for objects and not bodies which don't have fixed shape.
just curious, what is your general concept for action recognition here ?
I take in videos where a person is doing some activity (like walking, running, jumping, etc.) and I want the system to recognize which activity it is.
still, there's a huge, empty gap between your (generic)description above, and "use SVM with LBP histogram". care to explain more ?
I have normally seen the use of SVM for object detection. I just wanted to know if it can be used for action recognition as well, taking LBP histogram as the feature vector.
Sorry but I am new to this. It would be better if you ask specific questions. Thanks.
yoyo, no fear, all i'm trying, is to get you to explain more ;)
I am currently using 100 images for both classes for SVM training, and testing using one of those images only. But the output of SVM->predict, for every input, is always 0. Can it be because of less number of training images?