Abstarct: Abstract Today, through deep learning, many creations are made in most industries, and these creations have become a part of our daily lives. Deep learning has achieved many successes in various fields such as image processing, audio processing, and natural language processing. Also, especially in the field of recognizing and objects detective in the image, many achievements have been achieved by using convolutional neural networks. In this dissertation, we present a network for segmenting objects in images, baxsed on the Mask R-CNN. In the proposed network, first, the feature map of the input image is extracted by a deep neural network, then the important areas in the image are determined by a region proposed network. Depending on the output of the region proposed network, the correct areas are identified by a modified process and the objects in the image are detected. The pixels of the object will then be separated and labeled using a mask-generating network. The proposed method in general can be divided into 4 different parts. The various parts of the proposed network can be trained and in a global process using separate data, will train and test. According to the results of the network output, the proposed network is more accurate than the Mask R-CNN network and detects and separates the objects in the image well.