Advancing Facial Sketching and Recognition

Sketching the face is the essential step to identify any criminal. Till now we are using a traditional approach for sketching the face based on description given by witness person professional artist draws the sketch, but this method is very time consuming, and accuracy totally depends on the interpretation skills of artist. The novel approach used towards the forensic face sketching and recognition process is discussed in this research work. Also, this approach helps to overcome the limitations of traditional methods used for investigation process. The main objective of this work is to build a user-friendly tool that assists investigators to draw facial sketches which eliminates the need of a professional sketch artist and simplifies the investigation process. A drag-and-drop interface with predefined categories of facial features are used to develop the tool. To match the sketches with the existing databases, the Convolutional Neural Networks (CNNs) model is used. The system gives an option to colourize the sketches to make the recognition process efficient. Variational Autoencoders (VAE) are used to accurately colourize the sketches and give them a realistic image texture. The application helps in reducing the reliance of sketch artists by simplifying the sketching process and it requires no additional training to use the application.