High Resolution Panoramic Video Synthesis and Objects Detection on Panoramic Video

We propose an approach to the creation of a panorama viewport and objects detection within it in real-time on the base of the set of videos from the assembly of cameras. The task of the panorama viewport generation is significantly different from the panoramic image generation. We perform image stitching only with a part of cameras views (region of interest - ROI) and display the resulting images flow on the user's screen. The complexity of this problem is in processing the images from the set of cameras at the moment, since it is impossible to stitch, process and save the whole panorama in the user device memory. The main practical application of the proposed technology is video surveillance for transportation, industry and public objects (airports, railway stations, sea and river ports, plants, stadiums, etc.). Nowadays, there are many scientific articles exploring panorama stitching methods. However, not many existing algorithms can be used for the panorama stitching in real-time as is. Real-time panorama stitching is the value of the proposed approach.