The EDAG Group has developed software that enables the use of artificial intelligence functions for autonomous driving even in poor visibility conditions. The DiFoRem (Dirt & Fog Removal) system can compensate for visual impairments caused by dirt, fog or defective areas of the lens in real time.
In order to digitally remove physical contamination on a camera lens, the impaired areas of an incoming image are first algorithmically detected and tagged as such. During the training process, neural networks are trained to reconstruct the corrupted images by learning the relation between clean and impaired images. Artificial intelligence then reconstructs the contaminated areas.