This idea of the product is a Saas-Based CCTV surveillance platform with face recognition of car drivers. The product allows users to provide an IP Camera Link that differentiates users based on their face and car number plate and generates data based on the details. The breakthrough is that the system is also capable of giving alerts on the arrival of any specific person in the proximity.
To identify the faces inside the car the system needed a robust face recognition service through the CCTV. The bigger challenge was to have reasonable accuracy in a brief time since face recognition has to be real-time.
The client needs reports for the number of footsteps inside the premise along with the number of cars parked in the parking all this is combined with the In and Out reports by car number.
Since the client wants a scalable yet cost-effective ai-solution, we successfully developed a bespoke face recognition feature from scratch without using any third-party face recognition service like Amazon Rekognition.
We had to provide a solution that would track every incoming & Outgoing vehicle with a timestamp so we combined CCTV footage of the entry point with Computer vision to capture all the details of the car like make, model, color & number plate. This allowed us to gather information and create reports that the client earnestly wanted.
One of the most interesting requirements of the client was to detect people who may be threats as well as VIPs. Since the system is already equipped with a face recognition feature, we set up an alert feature. Every time a tagged person’s face is recognized by any of the surveillance systems, it will notify the admin about it.
83% Face recognition accuracy along with black male-female and white male-female demographics as per NIST’s Facial Recognition Vendor Test.
