The Unet architecture consists of contracting and expansion paths. The contracting part consists of a repeated block of 3x3 convolution and max-pooling layer. Whereas the expansion part consists of 2x2 upsampling block. Also, the features from the contracting part are concatenating with upsampling results to get high-resolution features.
- Python 3.8
- Numpy 1.19.2
- Matplotlib 3.3.2
- Tensorflow 2.4.1