Image recognition 101
In this workshop I would like to present introduction to image recognition using OpenCV. We would try to look for specific images on webpages screenshots using basic feature detection and matching algorithms.
I would like to prepare workshop regarding basics of image recognition with exercises prepared as jupyter notebook to avoid typing too much and make sure that all the participants are on the same page.
Firstly I would present how to start working with images in Python using PIL and OpenCV functions. I would like to show how easy it is to play around with those and have instantly some interesting, graphical results. Then I would like to explain basics of image recognition and show how to detect keypoints of an image using algorithms such as SIFT or ORB. That would give participants an idea of what image recognition actually is and how does it work. After that I want to show matching process itself using FLANN matcher. We would play around with different parameters of used algorithms to emphasise how important it is to tune them properly to achieve good results. That gives as space to talk about false positives and methods to avoid them. Lastly I would like to present simple solution to match image occurring multiple times in target image with algorithms that does not provide this feature out of the box. To achieve that I would like to introduce Mean Shift clustering algorithm implemented in scikit-learn package.
Break-out sessions are expected to last around 60 minutes. They will have limited seats and will have a sign-up option provided on the ticketholder website, once the schedule is known.