Defect detection in objects

asked 2020-08-10 12:38:57 -0600

FarshadM gravatar image

updated 2020-08-10 18:11:35 -0600

Hello friends,

I'm working on a AR project which I'm planning to develop on Hololens eventually, and I'm trying to find if the order of colors in a connector does not follow the reference object. To give an example, let's say the correct order is Yellow black black and red:

correct color order

and looking at this image with the wrong order, we should be able to tell the order of colors is wrong:

wrong color order

I'm open to use OpenCV or online learning algorithms, but I just wanted to know if anyone has an idea of a proper algorithm or approach to do so.

Thank you!

edit retag flag offensive close merge delete


please upload your images HERE, not on imgur (which is an absolute shitsite. blocked from several countries for a reason..)

berak gravatar imageberak ( 2020-08-10 14:03:50 -0600 )edit

then -- what have you tried, so far ? show us !

(we won't write your program !)

berak gravatar imageberak ( 2020-08-10 14:06:07 -0600 )edit

I have tried Microsoft custom vision and also dlib to label both positive and negative images, but when it comes to Hololens it's not much accurate. I was thinking to use semantic segmentation and I thought I ask my question here to know if somebody had similar problem before

I can write the program myself, I just want to know what the proper "approach", or "algorithm" is. I'm new to image processing and knowing the right approach for this, would be a great help.

FarshadM gravatar imageFarshadM ( 2020-08-10 18:14:35 -0600 )edit