We need someone that can program a rotation and scale invariant shape recognition algorithm for a OCR application. This algorithm should be implemented in C++ with OpenCV compilable for a linux OS (ubuntu).
The algorithm will receive as input an image and it will return the decoded number, for example:
Attached to the job posting is the image database to be used for training / testing purposes. Anyone who wants to apply for this job should be able code an algorithm that achieves at least a 95% accuracy. I strongly recommend you test your algorithms locally and only bid if you have an algorithm that will work.
HOW TO APPLY
To apply for this job you need to have either a very strong profile and feedback in computer vision / image processing or send attached the confusion matrix for the 9 classes in the image file obtained by using your algorithm (use any classification algorithm you like, we don't need this part implemented).
Please if you don't know what a confusion matrix is, it is best you do not apply.
computer vision, image processing, pattern recognition, ocr, c++, opencv