End to end testing including person detection via Rekognition. If the badge is green, it means we were able to encode & stream a sensor_msgs::msg::Image
video all the way to Kinesis Video, analyze the results in Amazon Rekognition and read them back.
Media pipeline flow:
ros2 bag play
-> h264_video_encoder
-> kinesis_video_streamer
-> Amazon Kinesis Video Streams -> Amazon Rekognition
Data flow (Rekognition results):
Amazon Rekognition -> Amazon Kinesis Data Streams -> kinesis_video_streamer
writes the results to topic /rekognition/results
To run locally on your development machine, clone this repository and invoke:
./run_integration_tests.sh
If running inside a minimal ROS2 environment (e.g. docker without wget, rosws, aws-cli etc.), or encountering missing dependencies issues - first run ./install_deps.sh
(with sudo if on your development machine).