Jan 13th-17th
- Started Deer Dissuader research
Thoughts on beginning deer dissuader project and project sketch:
The week of the 19th was not incredibly productive, I got the lowdown on the bird deterrent box and how the inside works. Lots of things need some reworking but I need to focus on getting a program to recognize a deer. I have limited knowledge of training an AI model but got helpful suggestions about OpenCV and yolov11. Found some tutorials and documentation, will start to implement these tools.
Some important tasks I need to do this week:
- Buy camera
- OpenCV and yolov11 documentation
- Break open the raspberry pi and look at existing code
Week 4 (1/26/25 – 2/1/25)
!! Solar Power due
| 1/26/25 | 1/27/25 | 1/28/25 | 1/29/25 | 1/30/25 | 1/31/25 | 2/1/25 | |
| To-do | – OpenCV & yolov11 research | – Expand dataset – Order camera? | – Expand dataset – Get solder and 3D printer certified 🤠 | – Order camera rain-check | – Open pi | – Expand dataset | – Work on exterior camouflage |
| Completed | – Opened roboflow workspace – uploaded photos to train yolov11 model | – Ordered camera, cable, ssd, and ssd adapter/enclosure | – Opened pi, spent 3 hours waiting for opencv to download (it didn’t work) |
Week 5 (2/2/25 – 2/8/25)
!! Exterior camouflage due
| 2/2/25 | 2/3/25 | 2/4/25 | 2/5/25 | 2/6/25 | 2/7/25 | 2/8/25 | |
| To-do | – Need to flash pi | – yolov5 and dependencies | – Set up camera and storage – Camouflage if no delivery | – Go over code | – Go over code | – Deploy? | |
| Completed | – Flashed pi and downloaded OpenCV!!! | – Downloading rest of software |
Week 4 had some bumps at the end. When I finally had time to look into the raspberry pi, I had trouble getting anything downloading with the desktop GUI being active. I believe it was taking up a lot of ram and made downloading very slow. What’s even worse is I think it was a 32 bit OS because certain dependencies were not downloading. Because of these hiccups, they will have to wait until this week, week 5. Additionally, working on camouflage will have to wait for this week too. I hope to get it done Tuesday or Wednesday before winter carnival.
I don’t have too many concerns for next week as long as I can get the model deployed this week. I am currently in the midst of re-flashing the pi and downloading the software we need. If all goes well we are back in business. If not, rip CDR.
Week 6 (2/9/25 – 2/15/25)
| 2/9/25 | 2/10/25 | 2/11/25 | 2/12/25 | 2/13/25 | 2/14/25 | 2/15/25 | |
| To-do | – Deploy if not done yet | – Test with images | – CDR!! | ||||
| Completed | – Obtained roboflow data – Found out roboflow costs to train model | – Research | – Set up data for yolov8 model | – Trained model 🙂 – CDR!! |
Hurdles were jumped in week 6, roboflow was a pain as we found out you have to pay to deploy or export a model they trained. However, you can take datasets from the website and the team already decided on deervision as it was high accuracy, plenty of data, and free to export. 🙂 I was able to use chatGPT and a bunch of guides to figure out how to get a python virtual environment to run yolov8. I have been going back and forth on yolo versions but 8 is the most recent stable version for the pi. I was very proud to get python to work and get the model trained. The next step to tackle is the camera driver issues. 🙁
Week 7 (2/16/25 – 2/22/25)
| 2/16/25 | 2/17/25 | 2/18/25 | 2/19/25 | 2/20/25 | 2/21/25 | 2/22/25 | |
| To-do | – First weekly meeting (5pm) | – Research camera issues and tinker | – Research camera issues and tinker | – Hopefully fix camera and move model to pi | – Organize pi | Spring Break | |
| Completed |
Week 7 was not very productive. The next week was spring break and I had James Burke take the raspberry pi’s to debug the camera. Since only 1 person could take the raspberry pi’s, I was left with not much to do that week. Week 8 will be more productive now that we are on campus, I would like to delegate responsibilities at the meeting on Monday and start debugging the camera and testing the model on the pi. Without a camera, I would like to test the model on my laptop. We may need to discuss an alternate route if the camera will not work.
Week 8 (3/2/25 – 3/8/25)
| 3/2/25 | 3/3/25 | 3/4/25 | 3/5/25 | 3/6/25 | 3/7/25 | 3/8/25 | |
| To-do | – Team meeting – Set goals for this week | – Camera | – Camera – Test model on laptop | – Organize pi and move model | – Move model rain-check – Test model | ||
| Completed | Camera Works!!! |
Week 8 was breakthrough as we finally got the camera to work. I also wrote a howto for programming the pi and got access to our github. github link: https://github.com/James07Burke/OSHE-Deer-Dissuader
Week 9 (3/9/25 – 3/15/25)
| 3/9/25 | 3/10/25 | 3/11/25 | 3/12/25 | 3/13/25 | 3/14/25 | 3/15/25 | |
| To-do | – Team meeting | – Upload to github | – Model move over? | – Present | – Test model performance on pi | ||
| Completed | – Discussed 3D models | – Measured Camera dimensions – Presented |
Camera works great, drivers are finicky and have only worked on Debian 11 so far. I need to get help from someone on how to analyze the drivers and see if there’s any way we can get it to work on Debian 12 because Debian 11’s EOL is in 2026. 🙁 Otherwise the week has not been very productive, I have been so busy I have not had time to sit and work on the pi. My weekends are when I’m most productive. To combat procrastination, I will leave it running on tech’s wifi and ssh in. 😛
Week 10 (3/16/25 – 3/22/25)
| 3/16/25 | 3/17/25 | 3/18/25 | 3/19/25 | 3/20/25 | 3/21/25 | 3/22/25 | |
| To-do | – Work on drivers – Work on github | – Team meeting | – Model move? – Watch some videos on python | ||||
| Completed | – Debian 12 camera drivers work | – Python coding started |
Last week was very big over the weekend, the arducam drivers were just recently updated to be usable in Debian 12 bookworm. This extends our os lifespan which is nice. Additionally, James and I started coding in python to use the model we created a few weeks ago. The model may need some more training over next week but once we get it reliably recognizing deer our project is essentially finished. I need to do more training and coding. We plan to get the github updated either thursday or over the weekend.
Week 11 (3/23/25 – 3/29/25)
| 3/23/25 | 3/24/25 | 3/25/25 | 3/26/25 | 3/27/25 | 3/28/25 | 3/29/25 | |
| To-do | – Team meeting | – Polish coding – Update Github | – Testing pi | ||||
| Completed | – Computer vision code – Testing started | – Updated github | – Wrote more code | – Wrote more code – Testing with laptop |
Week 11 was crunch time. I spent the whole week (10+ hours) working on python code and testing it with my laptop. Our model had to be trained again so we retrained it on James’s laptop as it had a little more power than mine. The weekend was when we fixed the code to do exactly as we wanted, it deploys the audio when a deer is detected and records the next 10 seconds and saves the video with a timestamp. I am very excited about the progress we have made. Switching to the pi, however, has a few issues as of right now and will need to be fixed before our presentation on Monday. Worse comes to worst I can present what the code does via my laptop. This may be the case if the speakers and pi cannot be hooked up together seamlessly yet.
Picture for final checkoff:
Week 11 (3/30/25 – 4/5/25)
| 3/30/25 | 3/31/25 | 4/1/25 | 4/2/25 | 4/3/25 | 4/4/25 | 4/5/25 | |
| To-do | – Finish python and move everything to pi | – Finish pi and put in case – Checkoff @ 6:30 | – Tweak pi and get up and running | ||||
| Completed | – Python complete, laptop tested – Moved to pi, code does not work on pi yet |