An object location for Brazilian boards, based in U-Net Segmentation using pre-trainded weights VGG16. For validation of the model, the Dice coefficient was used, in which it obtained 0.8088, which we can say that hit 80% of the expected location.
An object location for Brazilian boards, based in U-Net Segmentation using pre-trainded weights VGG16. For validation of the model, the Dice coefficient was used, in which it obtained 0.8088, which we can say that hit 80% of the expected location.