Pražské Pyvo #133

OpenCV is a standard image processing library for both classical and deep-learning computer vision. Using Python with OpenCV makes for a powerful tool to extract meaning from images and camera streams.

One can exploit a range of filtering and shape-detecting algorithms to create simple but reliable object detectors or trackers using only a few creative, hard-coded heuristics, without the need for extensive data collection and annotation. Also, the algorithms can achieve a much higher FPS than current state-of-the-art neural networks.

In this talk, we will first demonstrate the basic tools to capture a camera stream, then showcase filtering capabilities such as thresholding, gradients, edge detection and finally detect and track objects. There will be case studies of how to align documents lying on a desk or read barcodes from any object in real time. By the end of the talk you will have mastered a few pipelines to extract meaning from images and videos.