This project was one of my effort for project funding a research for our UAV team. We modified our EVARO UAV for amphibious terrain and mini PC system to retrieve video feed during flight and process for simple human detection.
This project was part of accepted Student Creativity Programme (Program Kreativitas Mahasiswa | PKM) 2022. I have to implement computer vision using mini PC using Raspberry Pi 3B to detect disaster victim as part of research topic.
I opted to use YoloV3-tiny for my computer vision, by computation resource Pi 3B won’t cut it alone therefore I had to use the C++ build and use small resolution input. For demonstration purpose we feed the video feed into the program and it will output the result with highlighted object of interest.
Here provided simplified flowchart of the system:
I provided several test results in form of video, due to time constraint and tight schedule, we only tests the UAV within university facility area.
Static Test: Horizontal Hover Test: Vertical Hover Test: