Abstract Visioner
Frontend description:
We used Angular and PrimeNG icons to create UI, images are downloaded directly from https://sachsen.museum-digital.de/series/309. User is able to select two images and send them to the server, then after certain amout of time the server returns a mixed image and the user can download it.
Backend description:
Technologies used: Python, Flask, Opencv, Pytorch, https://github.com/moritztng/prism, Pillow.
Data is download (like titles and descriptions) from https://sachsen.museum-digital.de/series/309 with BeautifulSoup Library and store it in a JSON file.
Server receives two images' URLs from a client, download it and performs blending operation (the algorithm is described on our Github).
CUDA 11.3 is highly recommended to use on the server. It sharply decreases a processing time.