0
$\begingroup$

This is the situation. I'm training a model to recognize letters of the Alphabet. There are 26 classes. When writing the code for 26 classes, and loading nearly 100,000 images to train, I'm having a lot of issues. However, I was able to successfully train the model to work on 10 letter increments. As in, A-J, K-T, and then U-Z. These three work perfectly fine (A-Z does not).

Question: Can I train my A-J model, save the H5. Then, train the K-T, and U-Z after and then MERGE the H5 files together? I understand it's possible to train all A-Z using fewer images and then retrain the model with a different image set, but the issues are coming when I'm doing a 26 class system - hence I'm asking if I can do it in increments of 10 and merge it after.

$\endgroup$
0
$\begingroup$

To get around this issue, I would suggest doing training in batches. This solves the issue of not being able to load 100,000 images and also removes the need to figure out how to merge the models.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.