Face Sketch Recognition Using Image Synthesis And Deep Learning

Criminal investigations heavily rely on crime scene sketches. To appropriately categorize sketch photos from crime scenes, this research suggests a deep learning based sketch image classification method. Convolutional Neural Network (CNN) is the algorithm utilized to classify each sketch image in the proposed deep learning based system. This method uses a custom image dataset that includes images of various people’s facial sketches. Moreover, the suggested system takes both color and sketch images. Several systems that automatically recognize subjects shown in sketches based on eyewitness descriptions have been developed recently however, when real-world forensic sketches are used their performance frequently suffers. Despite its effectiveness in many application domains, such as conventional face recognition use deep learning for face photo sketch recognition. This is primarily because there aren’t enough available sketch images to adequately train big networks. In order to address significant issues with the current face recognition system, a deep learning face synthesis and recognition system is proposed in this study. For face identification in the suggested system, the input can be either a color image or a sketch image. Images of various people’s face sketches will be utilized to train the recognition algorithm. The training algorithm is a deep neural network. The system may be able to recognize input face images once training is completed. This system comprises two steps for face recognition and synthesis. If the new input is a sketch image, the recognition stage receives the input right away. The GAN network is used to produce equivalent sketch images if the input is a color image. In this instance, the input to the CNN for classification is a synthesized image. Thus, the proposed system is a combination of image synthesis and deep learning.