Recent advances in imaging and robotics technology have been accompaniedythe introduction ofew methods and capabilities. Today, one way of imaging the gastrointestinal is to use wirelecapsule endoscopy. These capsules are capable of imaging various parts of the gastrointestinal and are preferredover other available methods due to their ease of use. Although these capsules have significant advantages, they also have disadvantages, such as limited battery life and bandwidth, low frame rate, and image quality. One way to overcome these problems is to compress the captured images inside the capsule, which is lossy, lossless, and near-lossless compression with a view of spatial redundancy. This dissertation aims to present a method with a hardware perspective to compress images in endoscopic capsules using deep neural networks. In this dissertation, two methods of compression are presented. In the first method, using a neural network, we perform align="left" Keywords: Capsule Endoscopy, Segmentation, Compression, Neural Network
