The main goal is to create a U-Net network from scratch by understanding the paper.

Further goals:

• Modification of the MobileNetV2 SSD for segmentation in realtime.
• Training of the UNet on Medical imaging datasets from Kaggle and the Traffic datasets.
• Other optimisations if possible

Paper link from https://arxiv.org/abs/1505.04597

• Code the network
• Dataset
• Training
• Results
• Video
• Explanation