heidelbaeren

In the course of the HIDA DATATHON for Grand Challenges on Climate Change 2020.

The challenge

Using high level open source frameworks for landscape image segmentation with PyTorch

Our approach

The configuration

architecture | pyramid attention network (PAN); encoder | se_resnext50_32x4d; batchsize | 8; learning rate | 0.0001; activation | sigmoid; data_size | 256 x 256

Used resources:

PyTorch: https://pytorch.org/

PyTorch Lightning: https://pytorchlightning.ai/

Segmentation models: https://github.com/qubvel/segmentation_models.pytorch

Our results

Validation segmentation DICE score: ~ 95%