Tolerate Failures of the Visual Camera With Robust Image Classifiers

Deep Neural Networks (DNNs) have become an enabling technology for building accurate image classifiers, and are increasingly being applied in many ICT systems such as autonomous vehicles. Unfortunately, classifiers can be deceived by images that altered due to failures of the visual camera, preventing proper execution classification process. Therefore, it is utmost importance build guarantee even presence camera failures. This study crafts robust augmenting training set with artificially simulate effects Such a data augmentation approach improves accuracy respect most common approaches, absence To provide experimental evidence our claims, we exercise three DNN on datasets, which inject into camera. Finally, eXplainable AI debate why trained proposed this tolerate