Michael’s Status Report for 4/20/2024

Weekly Accomplishments

1. Debugged issue where both the ribbon cable & raspberry pi CSI port were broken. Likely caused from transport (static electricity?)
2. Completed scanner
   1. Assembled full scanner unit
      1. Printed case v2 (allows for cable to be inserted)
      2. Soldered I2C pins to control LCD
      3. Assembled everything in to case
   2. Works best around 2FPS because higher framerates back up on the network
   3. Roughly 0.75s of camera latency
   4. Videos:
      Demo of Scanner Working @ 2FPS w/ 1s delay
      Demo of Scanner insert/remove working
3. Last week: Fixed server issues an now YOLOv8 runs on server
4. Started Final Presentation

Overall Progress

• Final Presentation needs a lot more work
• Custom YOLOv8 model needs to be retrained

Next Week’s Goals

• Finish final presentation slides (Sunday)
• Train YOLOv8 model
• Connect to database API

What I’ve Learned

I’ve learned how to use a variety of tools as a part of this project. For starters, I learned how to use both YOLOv8 and an Nvidia Jetson Nano. The Jeston was particularly tricky because the 2GB version available in the ECE inventory lost official support last year. A lot of the documentation was out of date, and so there was some modificiations that had to be be made to the instructions that I had to dig around online for. For YOLOv8 I was luckily introduced to YOLO in my Autonomous Robotics course this semester, but I learned most of the tools by scouring the example videos and documention.

While not a completely new skill, I had to learn how computer networks transfer data in much more depth than I had ever worked with before. There were a lot of good sources including a blog by a guy who wrote a library to make raspberry pi security cameras on his farm and the corresponding examples on his github. On top of that I learned about 0MQ which has a handshake like TCP but is faster like UDP in exchange for lossy data (which is fine for working with streaming images).