Hi,this is my capstone project(毕设项目) @sjtu ME F2002020 Juandan Luo
the algorithm is composed by 2 parts, the first part we use the yolov5 to do the segment task,the original code:https://github.com/z1069614715/yolov5-seg and the second part is wrote in the image_process_formal.py。
step1:
step2:
you can see the transforming detail in the step2.png (not the image above)
the label was done by using labelme。
we can not afford you the completed dataset(so sorry),but there is one best.py that we have trained , you can use that in the predict.py pr predict_test.py。
you can also contact [email protected] for more detatils