There are several deep learning-based algorithms that were proposed to solve image segmentation tasks. A sliding window approach can be applied at a pixel level for segmentation. A sliding window approach takes an image and breaks the image into smaller crops. Every crop of the image is classified for a label. This approach is expensive and inefficient because it doesn't reuse the shared features between the overlapping patches. In the following sections, we will discuss a few algorithms that can overcome this problem.
The Fully Convolutional Network (FCN) introduced the idea of an end-to-end convolutional network. Any standard CNN architecture can be used for FCN by removing the fully connected layers, and the implementation of the same was shown in Chapter 4, Object Detection. The fully connected layers are replaced by a convolution layer. The depth is higher in the final layers and the size is smaller. Hence, 1D convolution...