Random Shadows and Highlights: A new data augmentation method for extreme lighting conditions

Abstract

In this paper, we propose a new data augmentation method, Random Shadows and Highlights (RSH) to acquire robustness against lighting perturbations. Our method creates random shadows and highlights on images, thus challenging the neural network during the learning process such that it acquires immunity against such input corruptions in real world applications. It is a parameter-learning free method which can be integrated into most vision related learning applications effortlessly. With extensive experimentation, we demonstrate that RSH not only increases the robustness of the models against lighting perturbations, but also reduces over-fitting significantly. Thus RSH should be considered essential for all vision related learning systems. Code is available at: https://github.com/OsamaMazhar/ Random-Shadows-Highlights.

GEM: Glare or Gloom, I Can Still See You -- End-to-End Multimodal Object Detection

“…We employ ResNet-50 as the features extractor, while we also train g sc on MobileNet v2 [31] for the FLIR thermal dataset. To guide the fusion process and mimic harsh lighting condition for the RGB sensor, we also employ Random Shadows and Highlights (RSH) data augmentation as proposed in [7]. RSH develops immunity against lighting perturbations in the convolutional neural networks, which is desirable for real world applications.…”

“…One way to address this problem is to utilize a dataaugmentation strategy [6]. It refers to the technique of perturbing data without altering class labels, and has been proven to greatly improve model robustness and generalization performance [7]. Nevertheless, this is insufficient for the cases where the sensor fails to acquire the information The research presented in this article was carried out as part of the OpenDR project, which has received funding from the European Union's Horizon 2020 research and innovation programme under Grant Agreement No.…”