Repository navigation
Trained on 27,977 samples, these weights have achieved an accuracy in classification of 87.0000% (evaluation). This model expects the input to be the full binary represented as a 1024x1024 image.
The accuracy figure comes from a set of samples unique from the training dataset. See the README for more detailed statistics.
Full Changelog: v1.1.1...v1.2.0