The traditional testing of multimedia features requires a lot of manual work like visual checking for possible artifacts and incorrect functionality of the algorithms from captured images and videos. Furthermore, all test results which are based on human observation are inherently subjective being dependent on the individual testers. Subjective testing can lead to unwanted delays and higher costs during product development.
A good way to address this challenge is to use mathematical algorithms and computational methods to find anomalous behaviour from multimedia content. This provides objective, reliable and repeatable way of testing functionality and quality of the processing algorithms. Moreover, by attaching computational testing to automated test environment we can guarantee faster problem detection and higher quality of the products themselves.