2015-11-10 01:00:03 -0600 | commented answer | Random Forest predict_prob in OpenCV 3.0 But instead of the pure class value I would like to have the probability of the sample(s) being part of the first class, as long as it is a 2-Class problem. In recent version we had Random Forests' predict_prob method to do so which I am missing in latest version. |
2015-11-09 01:53:46 -0600 | asked a question | Random Forest predict_prob in OpenCV 3.0 Hi, I'm looking for the predict probability functionality for binary classification task present in 2.x releases. As far as I see it is gone, is there a tendency towards getting it back? Best regards, chris |
2015-10-01 06:22:45 -0600 | received badge | ● Enthusiast |
2015-09-18 08:14:34 -0600 | received badge | ● Nice Question (source) |
2015-09-18 04:59:30 -0600 | commented question | SVM TrainAuto test_samples Looks like =) |
2015-09-18 02:59:36 -0600 | received badge | ● Student (source) |
2015-09-18 02:57:29 -0600 | asked a question | SVM TrainAuto test_samples Hi, I was looking for updating to 3.0 and checked the new TrainData class and how it works with the SVM TrainAuto. As far as I can see during cross validation, the temp_test_samples matrix is never filled with data or am I getting anything wrong? Mat temp_test_samples(test_sample_count, var_count, CV_32F); predict(temp_test_samples, temp_test_responses, 0);
Best regards, chris |