Huaizu Jiang
Huaizu Jiang
Actually, I got evaluation results in the log file when training the model. But there are "raw" and "smooth" results. Which one shall I use? Thanks.
It means you do not have the file "train.txt", which specifies names of training images.
I am not sure if I fully understand your question. The pairs you mentioned used to learn a binary classifier, indicating if two adjacent superpixels belonging to the same class....
It's a regression forest. The information gain is usually used for a classification forest.