Over the Winter break of 2023-2024, I wanted to do a project to gain experience in two areas I was interested in: TinyML and AWS. To do this, I decided to showcase my skills in a Live Computer Vision system, which is a field that I absolutely love.

I made a github repository for this project here:

For this project, I connected an OV7670 camera to an Arduino Nano 33 BLE Sense in order to create an effective Edge deployment system to capture live images of Saltine crackers. Then, I made a system on AWS consisting of an API gateway, Lambda, S3, and Sagemaker to create an effective cloud-based backend for storing images and conducting MLOps. All of this is summarized in this System Diagram I made.

Overall, the most fun part of the project was the research required to complete the entire project. You might be asking: how does someone choose to use an OV7670 camera or an Arduino Nano 33 BLE Sense for the hardware, or how does someone choose to use S3 or a Lambda in the backend? How do you choose what tools to use out of all possible options to choose from? Well, the answer to that is a ton of research poring over forums to determine what tools work well with what other tools. Furthermore, this was my first time creating a comprehensive AWS backend, including using Sagemaker, and learning how to do this required an absolute ton of watching AWS tutorials and just learning how everything works. For example, for roughly 80% of the project, I was planning on using an RDS storage in AWS to store metadata on the images. However, after performing more research, I found than an RDS was actually a much more expensive option that the one I needed, and after testing it out, I realized that I didn’t actually need it.

I’m really proud of this project because it required extensive research and the use of a professional tool (AWS), just like I would be doing at a job in the future. Overall, this project simply served as reinforcement that I have a ton of fun doing stuff like this, and I look forward to making a career out of this in the future.

I also made a cute picture below of the setup, I personally find this image really funny.