Graphics processing unit (GPU), although a powerful performance-booster, also has many security vulnerabilities. Due to these, the GPU can act as a safe-haven for stealthy malware and the weakest 'link' in the security 'chain'. In this paper, we present a survey of techniques for analyzing and improving GPU security. We classify the works on key attributes to highlight their similarities and differences. More than informing users and researchers about GPU security techniques, this survey aims to increase their awareness about GPU security vulnerabilities and potential countermeasures.
Multi-camera networks are becoming ubiquitous in a variety of applications related to medical imaging, education, entertainment, autonomous vehicles, civil security, defense etc. The foremost task in deploying a multi-camera network is camera calibration, which usually involves introducing an object with known geometry into the scene. However, most of the aforementioned applications necessitate non-intrusive automatic camera calibration. To this end, a class of camera auto-calibration methods imposes constraints on the camera network rather than on the scene. In particular, the inclusion of stereo cameras in a multi-camera network is known to improve calibration accuracy and preserve scale. Yet most of the methods relying on stereo cameras use custom-made stereo pairs, and such stereo pairs can definitely be considered imperfect; while the baseline distance can be fixed, one cannot guarantee the optical axes of two cameras to be parallel in such cases. In this paper, we propose a characterization of the imperfections in those stereo pairs with the assumption that such imperfections are within a considerably small, reasonable deviation range from the ideal values. Once the imperfections are quantified, we use an auto-calibration method to calibrate a set of stereo cameras. We provide a comparison of these results with those obtained under parallel optical axes assumption. The paper also reports results obtained from the utilization of synthetic visual data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.