Modern day’s criminals are getting smarter and with the use latest technologies they remain undetected during their crime, they make sure not to leave any biological evidence such as fingerprint or any kind of DNA samples. It makes the process of identifying criminals lengthy and difficult. A quick and easy solution to this problem would be considering facial recognition system. Now days with advanced development in technology CCTV cameras can be found installed in most buildings and even at traffic control areas for surveillance purpose. Later on footages from these cameras can be used for identifying individuals they can be criminals or suspects, missing persons and more. This proposed paper illustrates how to develop a sophisticated criminal identification system using Machine Learning and training artificial neural networks.