Server-side Model Training for Development Workbench

[akhil-rana]

Works with caMicroscope #426

Description

• The dataset zip is sent from the caMicroscope frontend and extracted into /dataset/<randomUserFolder>/
• The spritesheet and binary label files are parsed in /handlers/datasetHandler.js
• The model details i.e custom layers and parameters are handled at /handlers/modelTrainer.js along with model training process and some other utilities.
• Dockerfile has been updated with node:14-stretch-slim debian image since alpine has compatibility issues with TensorFlow C binaries. (Mentioned here)
• I wasn't able to test GPU training since I don't have nvidia (CUDA enabled) gpu; though it should work out of the box. If you are running on a Linux system that is CUDA compatible, try installing the GPU package using npm install @tensorflow/tfjs-node-gpu and replacing the require statement with require('@tensorflow/tfjs-node-gpu') in modelTrainer.js Note that proper drivers needs to installed on both base machine and the conatiner in order to use it.

How Has This Been Tested?

Tested on Firefox (79.0), Chrome (84.0), Brave (1.10.x) on Windows 10 and Linux-LTS (arch - deepin)

Types of changes

• [ ] Bug fix (non-breaking change which fixes an issue)
• [x] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)

Checklist:

• [x] My code follows the code style of this project.
• [ ] My change requires a change to the documentation.
• [ ] I have updated the documentation accordingly.

[akhil-rana]

I've tried maintaining the code style up to some extent. Do let me know of any suggestions or fixes if required. :)

[akhil-rana]

Should we have user permissions for accessing the 'server-side training' feature ?? Like only Admin and Editor maybe.

[akhil-rana]

Okay. Thanks. I'll add the permissions.