v1.0 trained on 130+ high res images. (3990 steps)
v2.0 trained on additional new 168 images (8770 steps)
Will add more images to the dataset and keep training the model for better results.