Boundary detection using Machine/Deep learning

I have the following x and y points that make a shape, however the points are not in order and when using a line plot all the lines intersect and move randomly. Is there a way to use machine/deep learning to order the points so that the shapes are accurately drawn.

[Ahmed Abdulla]

This is how the points are ordered when i use a line plot

[Simon Chan]

Did you try function boundary and polyshape?

[KSSV]

With some tweaking in the code given here: https://in.mathworks.com/matlabcentral/answers/1725865-order-points-for-plotting?s_tid=prof_contriblnk you can achieve what you want.

[Ahmed Abdulla]

@Simon Chan Yes i tried both but the results are never exact and not all points get used :(

[Ahmed Abdulla]

@KSSV I have tried tweaking the codes, however depending on which random point we start in the code have minor failures and doesnt generalise well to all shapes :(

[KSSV]

I have multiple questions on this contest.

1. Where and how are you generating these figures/ points/ data?
2. Can't you save the points in order, while generating?
3. How many shapes are there?
4. For what shapes, the code fails (minor failure)?
5. For how many shapes the code worked?

[Ahmed Abdulla]

@KSSV

1. The points are generated from an STL mesh where the points are imported into MATLAB
2. I have looked into this but unfortunately that wasnt possible
3. I would like the code to work for any shape given to it
4. Where the closest point interms of distance isnt the logical position to transition to

[KSSV]

Attach your stl file.

You have not answered how and where this stl file is generated? Point 4 is not clear to me.