- I'll upload (opencv tutorial code + comments) and my project using opencv.
- This project will be written with C++. and you should use hsv_ransac_ver-0.cpp.
- rest of cpp files are my trial and errors, so i don't recommand you to use them.
- this project's strategy is using lane's color(which is typically white and yellow) transformed to hsvColor.
- after i croped lane's points with color strategy, I will use ransac algorithm.
check original img and result img.
- gather lane's points
- left_lane ransac
- full ransac and find_route to drive through
- caution: you may be need to edit roi function's vertices array in _.cpp, reason followed under this line.
ransac malfunctions if input img contains strange object which is not road. (but it typially works.)
- like these errors
- This program's input is still img. I will revise it so this project use video as a input.
- hsv_ransac_ver-0.cpp will be refactorzied soon.
-
change format to class to make code more shorter.
-
it has two version of ransac(1차함수, 3차함수) and choose one of them by c_max value, but i will chage it soon.
- so among those two error, i corrected second one so far...
-
I will construct YOLO network in C++, to make object detector in video files.
-
Not only this, but also I will check my model's throughput and measure workloads.