For developing new technique to identify forged signs: Centre grants copyright to PU

The new model will help identify forged and real signatures in forensic examinations. The model was developed using a deep learning technique called convolutional neural network (CNN) to identify questioned signatures. It was trained directly on 1,400 images of signatures, where 700 belong to real signatures while the other 700 belong to forged ones. Out of 1,400 images, the software can determine whether the signature is real or forged with 84.5 per cent accuracy. This technique has the potential to identify signatures under investigation in forensic science laboratories/institutes. It can aid other questioned document examiners in identifying signatures in less time and with greater accuracy.

Professor Krishan, who is a renowned forensic scientist, started working on the applications of AI in forensics in early 2024 with his research team. The main goal of the research team is to develop AI software/models that may assist forensic research and crime scene investigation.