Create a semantic segmentation model using Tensorflow that determines the region of
the disease looking at the 3D medical image provided. Only task to be done is colon
cancer segmentation.
1. Utilize Transfer learning to train the model faster
2. Train and validate the model on 2D images (converted from 3D)
3. Demonstrate by inferencing on both 2D images and 3D images.
4. Keep the latency of the model (forward pass) less than 250ms on the GPU
provided on Colab (Nvidia T4)
Hint: Use “nibabel” library available using pip for data preparation.