During this week, my main focus was on setting up the web streaming framework for the Raspberry Pis. The new web streaming program utilized the Flask web framework to grab camera frames from the camera connected to the Raspberry Pi, and send them to a local http server. Then, ngrok creates a tunnel between the local http port and a web server, so that the video stream is accessible wirelessly. After installing this system on our Raspberry Pi, we were able to use a Google Colab notebook to record a video from the Raspberry Pi camera directly to Google Drive, with a latency of under 10 seconds for the entirety of the recording.
My portion of the project remains on schedule. My upcoming objective is to continue integrating our web streaming framework with our CV backend. This will involve conducting further live tests with multiple web tunnels from multiple Raspberry Pis, in order to determine if we can properly synchronize the web streams of the main camera + the verification cameras. This could prove to be challenging given the increased latency and variation in network quality inherent in migrating our product’s communication system onto the cloud, but will need to be done in order to obtain accurate estimation and prediction results . A goal for next week is to record a test session with multiple cameras and our CV unit on the cloud.