Projektnummer

0062831

Zusammenfassung

During the last few years artificial neural networks, and in particular so-called deep neural networks, have made a huge leap forward in the field of computer vision. These methods provide spectacular results in object recognition and in the analysis of complex scenes. However, one of the main requirements for their application is that images used for input are in focus and sharp. Out of focus images appear blurry and prevent a successful analysis. In comparison our eyes are able to adjust the focus via accommodation. Is it possible to teach an artificial neural network depth perceptio...