Feb 2 Update

• Added in Yolo V3 object recognition model in Django backend for generating object tags.

• Checked in with Problems Lab and confirmed that we will get an office space. Need to wait for a few weeks still.

• Received Realsense Camera Model, researched setting it up with Raspberry Pi we had.

• Found out that raspberry pi is not the right platform for Intel’s Realsense camera model which need X86 architecture.

• We compared multiple different development platforms. First selected Intel Galalio board and tried setting up the Realsense SDK on it but eventually found it is not the best platform since it is not having enough bandwidth for data transmission, do not have GUI interface which is hard to config. And have very little amount of support on internet.

• After experimenting with Galalio board, we eventually selected Nvidia Jetson nano and purchased it from amazon.

• Compared between GCP and AWS’s storage service, went with GCP because “Cloud Storage archive class” is the cheapest option for storing large data for the long term. Which pricing will start at $0.0012 per GB per month ($1.23 per TB per month) when it launches later this year. That’s significantly cheaper than Microsoft’s Azure Cool Blob Storage, which costs $0.002 per GB per month and competitive with Amazon S3 Glacier, which is priced at $0.004 per GB per month.

Setup the GCP server VM with following specs:

Name: instance-1

Ubuntu 18.04

Storage 150GB

$30/mo

• Confirmed software stack: The software system of Guard is separated into two distinct stacks, one which operates onboard in the data collection vehicle, and one which operates in a cloud hosted backend (AWS EC2 or GCS). The onboard integrated software stack handles tasks such as sensor interfacing and data aggregation, while the backend stack handles Guard’s data annotating (perception), object storage, video manipulation (editing), and public API services for the end-user.