Video-Rekognition

A project uses AWS Rekognition to do video analysis.

• Re-implemented the front-end based on my previous implementation Smart Retailer;

AWS Setup

Take a look at this file for the backend, follow the format and use your own customized contents.

https://github.com/denven/Video-Rekognition/blob/master/back-end/.env.example

What business analytics do we get from videos?

The app will extrapolate simple business analytics from videos for the day to day operations.

• Number of customers in video
• Age, sex, emotions (physical)
• Average time in video
• Recurrences of previously analyzed people in videos
• Average time before recurrences

Screenshots

How to use this APP?

1. Hit Upload Tab at the SiderBar
2. Drag and drop mp4 files (files named as: VID_YYYYMMDD_HHMMSS.mp4 will be accepted)
3. Upload file and drink a cup of coffee to wait until the analysis is done!

Note:

• The video analysis takes time, and the waiting time varies from duration, motions, persons, file sizes, resolutions of the video. It will awalys take several minutes to for a 20s video;
• You should have your AWS account configured, including users/roles/credential keys, Rekognition, S3, SQS, SNS services setup etc.
• Be aware of your bill if you stick to uploading large size and complex videos for analysis.

What's the stack

Backend

• Node.js
• Express.js
• Postgres
• AWS Rekognition, S3, SNS, SQS
• Server Sent Events

Frontend

• React.js
• Material UI
• Material table
• reCharts