Face Recognition through Data Science
Our client is a Government Security Organization in India and has been ensuring a secure environment for the public and also keeping a track of all listed criminals, and suspects etc. As part of one of their process, they want to get rid of a face recognition process that is fully manual and prone to error. Muoro’s team in collaboration with the client implemented a solution so that this entire process could be automated, quick and free from any error.
Client has large amount of data related to people and their face images in their records. They wanted to setup an automated verification process for a certain section of people that is quick and fault proof. They also want to see a automated report by the end of month for the verified cases.
The overall business challenge was to make the process fully automated for face recognition and thereby reducing the time to service and cost involved in the complete process.
Muoro team was engaged with the client to automate the entire face recognition process with an automated report generation by the month end. To meet the growing technological demands and to improve the operational
efficiency of the process of the client, Muoro’s team implemented the following:
• Build a deep learning model that took a lot of sample face images and train model.
• Data of the client that consisted of face images with their names were feed into the system and model was re-trained fro these cases.
• The model was improved continuously so as to detect faces even in case of low light, a small object in-front of face etc.
• Lastly, as a outcome of implementation, a face image was being fed into this model at real time and the recognition done.
Neural Network in Deep learning: