pyronear / pyro-vision

Computer vision library for wildfire detection 🌲 Deep learning models in PyTorch & ONNX for inference on edge devices (e.g. Raspberry Pi)

Home Page:https://pyronear.org/pyro-vision/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[references] Add unittests or CI checks for training scripts on reduced dataset

blenzi opened this issue Β· comments

πŸš€ Feature

Add unittests or CI checks for training scripts

Motivation

Currently one has to test the scripts manually. The dataset is too large to allow a complete test so one idea is to do the training on a reduced one, to be defined.

Pitch

Alternatives

Additional context

I suggest considering having a trainer class like in https://github.com/frgfm/Holocron/tree/master/holocron/trainer, which can be unittested but keep the training script untested simply to avoid CI minutes running away too fast!

Closing this for now as we're gonna change the repo general setup shortly, feel free to reopen if you think it's still relevant!