An AI-enabled
Video Banking Platform
Video Banking Platform
Face match for customer identity verification using Amazon Rekognition
Match customers live photo, existing photo or document ID photo with each other and get a % match result. Configure the % threshold as per internal policy to accept or reject the customer’s case.Facial analysis using Amazon Rekognition
Real time analysis of the customer’s face to know gender, approx. age, and if customer is wearing any glasses, cap, facial hair etc.
Check customer liveness using Amazon Rekognition
Instant and seamless service to know if it’s the customer on the call or its pic in video or video in video
Realtime Deepfake detection using Amazon Rekognition
Detect deepfake video on live video call and get a % possibility of Liveness ranging from 0% to 100%.
Realtime speech to text using Amazon Transcribe
Convert the complete video interaction from speech to text with user specific comments, and timeline.
Language translation for over 70+ languages using Amazon Translate
- Real time translation of one user’s speech into another users language text.
- Complete translation of the call transcript into any other desired language.
Customer intent understanding
Analyzing the customer’s speech to understand the intent and decoding that into a usable instruction or service request for the bank.Automated video analysis
Analyzing the video to match answers to profile data, multiple people on call, multiple voices on call, customer wearing cap or glasses, eyes open or closed.
Additional AI models used
- Open AI Whisper Large V3 model custom finetuned with real video call data of over 100 hours
- Yolo 11 finetuned for ID, signature and multiple people detection
- BERT model for similarity string matching
- AWS OCR for ID card verification
- Whisper X model for accurate timeline detection