Seed Particle Tracking Computer Vision Project
In my Master's program, I built a seed tracking system using Computer Vision techniques for the company John Deere under the supervision of a Ph. D. student. The system was able to real-time track the trajectories of moving particles in a 3D landscape.
Here is a of the program tracking the movement of the centroids of the rice pieces moving through the air. This was also successful with the tracking of pellets and grass.
​
When running through the program, the code determined the change in 3D position as this took place.
The image above is a graphical representation of a Charuco Board in the focal region and the four cameras on the filming set-up.
​
The cameras were calibrated using the CV2 Python Repo and Charuco Boards. Using this method, the program was able to determine the locations of each camera in reference to each other and the focal region.
The camera set-up was built using t-rails and a PLA camera mounts that I designed on NX.
​
These allowed for me to rotate the HT-3000 Emergent Cameras about 2-axes of rotation. A close-up of the mount is shown above.