A Deep Learning-Based Approach for Identification and Recognition of Criminals

Face sketch recognition is one of the most researched issues in forensic science. Automatically retrieving suspect mug-shot images, police record can facilitate them swiftly tapered down and eliminate prospective suspects, but in most circumstances, a suspect's photographic image is not available. Sketching from the memories of an eyewitness or a victim is frequently the best substitute. In general, this procedure is slow and ineffective, as it does not allow for the identification and arrest of the appropriate culprit. As a result, a more powerful algorithm for even partial face sketch recognition is frequently beneficial. Many solutions have been offered in this scenario, particularly the techniques used in face recognition systems, which are regarded among the best and most effective. Our project uses deep learning and cloud infrastructure to allow users to create composite face sketches of suspects without the assistance of forensic artists using the application's drag and drop feature, and to automatically match the drawn composite face sketch with the police database much faster and more efficiently.