Live object detection prototype is built on AWS cloud using Rekognition service with LTE network. A Video is sent to AWS cloud and objects in the video are detected. Result video is streamed by RTMP.

1. Video is captured using RTMP camera app in the smartphone

2. Video frames are pushed to AWS EC2 RTMP VM using RTMP protocol on LTE network

3. In AWS EC2 VM, RTMP frames are captured to images by OpenCV VideoCapture function

4. AWS Rekognition object detection API’s are called and AWS Rekognition returns bounding boxes

5. Bounding boxes are drawn on each image

6. Images with bounding boxes are converted to video and are sent to the streaming VM using FFMPEG

7. Result video is streamed using RTMP protocol on the client