Amazon Rekognition provides highly accurate facial analysis and facial search capabilities that can be used to detect, analyze, and compare faces for facial verification, attendance counting, and more.
Lele (Guangzhou) Technology Development Co., Ltd. is a company engaged in the research, development, and sales of maternal and infant products. The company operates high-end diapers, baby skincare products, and other maternal and infant products, gaining consistent love and praise from young mothers in China. Since the company’s employees are located across the country, they need a centralized facial recognition attendance system.
(1) An Android camera application calls the AWS Kinesis Video Streams Producer API to push real-time video into an AWS Kinesis Video Stream.
(2) A Lambda function extracts key frames from the Kinesis Video Stream and sends them to the Amazon Rekognition service for facial detection and recognition.
(3) If a key frame is detected to have a face that matches one in a pre-defined collection, the frame image is stored in an S3 bucket, and its metadata (including the S3 path) is saved in a DynamoDB table.
(4) Attendance administrators query DynamoDB via a web UI fronted by the AWS API Gateway. The UI displays the returned images of employees who have clocked in.