Hi Raza,
UNet++ is an encoder-decoder based network for segmentation which consists of many layers like convolutional layers, batch normalization layer, concatenation layer, pooling layer, activation function etc. These layers are present in deep learning toolbox. You can refer to this documentation page for more information on how to use these layers. The architecture of UNet++ can be defined as stated in the paper of UNet++ using these layers. You can refer this custom example, which shows how to define a CNN architecture. Hope it will help.
