In the ML assisted film post production project, we aim to develop new tools to increase efficiency and quality in film and media post production by applying state-of-the-art ML technologies, semantic databases and search algorithms on compressed image files.
International project supported by Innosuisse: 109.856 INT-ICT (2024-2025)
In all three use cases that we focus on, various techniques from computer vision are required. From object detection, over image manipulation to image generation, in each of these fields, deep learning made an extraordinary leap and lasting impact. However, many companies are still struggling to bring the developed models to production. Usually, it is because the quality required for production is not sufficient, and the industry is not willingly to compromise efficiency for quality, especially because no one wants to be liable for the mistakes and correcting the mistakes is usually more expensive than performing the manual task. Nevertheless, there are attempts to introduce ML into film post-production.