INTELLIGENT SECURITY IN THE MARITIME SECTOR THROUGH ARTIFICIAL INTELLIGENCE-BASED GAIT ANALYSIS

Abstract

Maritime activities, including global shipping and port operations, rely on interconnected systems where access control is critical for cybersecurity. Gait-based recognition can provide a reliable biometric solution to enhance secure authentication in such maritime environments. This paper seeks to investigate gait recognition as an application of security deployed on deep learning, namely Convolutional Neural Networks (CNN) on gait images having four walking conditions of the CASIA-C dataset. Gait recognition is a promising type of biometric, especially in the maritime field as it can be applied in the identification of people and safeguarding security in restricted or sensitive regions. Our proposed technique proves the applicability of deep learning methods to increase the gait recognition performance, in particular, the performance of CNN versus the MobileNetV2. MobileNetV2 was employed over the CASIA-C dataset, results in 88% accuracy. Nonetheless, we performed a lot better because our results with CNN model reached a score of 92.79% accuracy and 87.33% precision. These results indicate that CNN design more valid and well-developed gait classification model in using security scenarios and especially in detecting individuals in different walking parameters.