how to train on my own database?

[manvirvirk]

how can i train on my own database? can anyone help?

[metobom]

I created a repo for this. Might help: https://github.com/metobom/prepare-data-for-UNet

[lyc1995452-star]

I created a repo for this. Might help: https://github.com/metobom/prepare-data-for-UNet

sorry, I have visit you put forwar the website, but I cannot understand your meanning.

[metobom]

I created a repo for this. Might help: https://github.com/metobom/prepare-data-for-UNet

sorry, I have visit you put forwar the website, but I cannot understand your meanning.

Which part you couldn't understand? I can answer your questions.

[lyc1995452-star]

I created a repo for this. Might help: https://github.com/metobom/prepare-data-for-UNet

sorry, I have visit you put forwar the website, but I cannot understand your meanning.

Which part you couldn't understand? I can answer your questions.

I am doing multi-classification and segmentation. I want to ask you how to conduct the training data. Is the 8-bit grayscale feasible? Can you explain it in detail?

[metobom]

I created a repo for this. Might help: https://github.com/metobom/prepare-data-for-UNet

sorry, I have visit you put forwar the website, but I cannot understand your meanning.

Which part you couldn't understand? I can answer your questions.

I am doing multi-classification and segmentation. I want to ask you how to conduct the training data. Is the 8-bit grayscale feasible? Can you explain it in detail?

Sorry I only did binary classification. But you can take a look at "adjustData" function inside of data.py. Also in "trainGenerator" function flag_multi_class is initially False. You can turn it true, put your data to required paths and try a small training.

[manvirvirk]

i have small database of RGB medical retina images and i want to segment it using UNET. How to do it using this repo??

[manvirvirk]

Actually i have generated more dilated images and want to segment those. Please help. THANKS