Closed madhavsund closed 8 years ago
We are trying eesen for OCR. We have a dataset with labels marked 1 to 320 . but some of the labels are not seen in training and testing set.only 218 labels are seen. the question is what would be the number of output neurons either 218 or 320?
You should use 320 labels
thank you... it works. my network is two layer with 320 neurons, but the convergence is very slow. what could be the reasons
the problem was with parameters... now it is working
We are trying eesen for OCR. We have a dataset with labels marked 1 to 320 . but some of the labels are not seen in training and testing set.only 218 labels are seen. the question is what would be the number of output neurons either 218 or 320?