UFOSend is seeking a computer vision/image processing engineer to build a software solution to compare two identical products captured in two different but controlled environments.
The products in this case are watch dials. The software solution will be looking for missing components, incorrect spellings, color mismatches, etc.
The software system should accept an input set of images to serve as the ground-truth, ideal images.
Then, on the assembly line, new images will be acquired using cameras. The products on the assembly line need to be compared to their ground-truth images, looking for defects, manufacturing problems, etc.
Below you can find examples of the ground-truth dials:
Along with examples of defects:
The end product should include:
- Source code with clear comments in the Python programming language
- A statement describing which algorithms were used and why the algorithm chosen is better than alternative approaches
- A clear statement of assumptions made for this project and when/where the implemented algorithm will not work
If at all possible, the software system should also highlight where in the image a defect occurs (not just the fact that one happened). This is a secondary goal of the software and while not a requirement, would be good to have.
To apply for this job please use the form below.