/Image_segmentation_with_U-net

Image segmentation using vanilla Unets with batch normalization.

Primary LanguageJupyter Notebook

Image_segmentation_with_U-net

Image segmentation using vanilla Unets with batch normalization layers after convolutional blocks. For this project, I have used Vanilla U-net architecture from the original paper [arXiv:1505.04597 [cs.CV]] and I have used images of motorbikes from [http://host.robots.ox.ac.uk/pascal/VOC/voc2012/] Due to very small resultant dataset, I have used albumentations library for Image augmentation and data generation. I have achieved 0.04 binary-crossentropy loss on the evaluation set.