Home

Smart security camera with Raspberry Pi Zero and AWS

The instructions to install a security camera are given in the following steps:

1. Install Motion and get it running automatically on Raspberry Pi. https://github.com/salekd/rpizero_smart_camera/wiki/Motion
2. Configure a palette of AWS Services for the workflow described below. https://github.com/salekd/rpizero_smart_camera/wiki/AWS-Services
   • Store images in Amazon S3.
   • Trigger an Amazon Lambda function when a new image is uploaded, taking care of the following steps.
   • The lambda function uses Amazon Rekognition to identify objects in the image and
   • sends e-mail notification using Amazon Simple Email Service if a human appears in the image.
3. Enable access to AWS from Raspberry pi. https://github.com/salekd/rpizero_smart_camera/wiki/AWS-on-Raspberry-Pi
4. Alternatively, TensorFlow Object Detection API can be used in Amazon Lambda. https://github.com/salekd/rpizero_smart_camera/wiki/TensorFlow-Object-Detection-API

The performance of both solutions is briefly discussed in https://github.com/salekd/rpizero_smart_camera/wiki/Performance

The AWS Services are not free of charge. The costs of both alternatives (Amazon Lambda with Amazon Rekognition and Amazon Lambda with TensorFlow Object Detection API) are described in section https://github.com/salekd/rpizero_smart_camera/wiki/Costs

I got the idea of improving a Raspberry Pi security camera with the Amazon Rekognition service from Mark West. Here are the original sources:

• https://github.com/markwest1972/smart-security-camera
• https://www.bouvet.no/bouvet-deler/utbrudd/building-a-motion-activated-security-camera-with-the-raspberry-pi-zero
• https://www.bouvet.no/bouvet-deler/utbrudd/smarten-up-your-pi-zero-web-camera-with-image-analysis-and-amazon-web-services-part-1

Read here about the release of TensorFlow Object Detection API https://opensource.googleblog.com/2017/06/supercharge-your-computer-vision-models.html and check out this amazing demo by Dat Tran https://medium.com/towards-data-science/building-a-real-time-object-recognition-app-with-tensorflow-and-opencv-b7a2b4ebdc32