Using DessedDilated with custom dataset

[SrijithBalachander]

Hey authors,

I was wondering how I could transfer the dessedDilated model to train with my pipeline using my custom datasets and dataloader. I have tried looking through files in the model directory, the settings files, _process.py etc. and I seem to be getting somewhere but its still not clear to me. What changes do I have to make to simply use the dessedDilated model?

Thanks

[dr-costas]

Hey!

I can see two ways of treating this. The first is by using a custom data loader (with the appropriate dataset scripts), and the second by getting the files for the dessed_dilated and use them in your own system.

---

To do the first one, you have to modify the file experiments/_processes.py, at lines 257 to 260:

with InformAboutProcess('Creating training data loader'):
        training_data = get_tut_sed_data_loader(
            split='training',
            **settings['data_loader'])

and at lines 262 to 265:

with InformAboutProcess('Creating validation data loader'):
        validation_data = get_tut_sed_data_loader(
            split='validation',
            **settings['data_loader'])

you will have to replace the function get_tut_sed_data_loader with your own one.

If you have special arguments, you can add those to the .yaml file with the settings that you want to use, under the key data_loader.

---

To do the second, you have to get the files:

• models/dessed_dilated.py,
• models/_modules/dessed_dnn.py, and
• models/_modules/depthwise_separable_conv_block.py

and use these files in your own code.

---

Let me know if the above worked!

Cheers!

[dr-costas]

Hey,

Have you been able to use the code with your own dataloader? Shall I close the issue?

[SrijithBalachander]

Hi,

Yes I was able to run the model on my own system. Got it working just an hour ago. Thank you. You can close the issue.

[dr-costas]

Happy to hear! I'm closing the issue.