Page 86 - tech fest 2025 אשדוד חוברת תקצירים

86
Airborne Facial Recognition System
ME-A-03
May Swisa; mayswisa0@gmail.com Refael Bareket; nrb2025@gmail.com
Advisors: Dr. Shayke Bilu1, Mr. Daniel Stoller1 1SCE - Shamoon College of Engineering, Ashdod
This research presents an airborne biometric identification system utilizing ‘convolutional neural networks’ (CNNs) for real-time facial recognition in dynamic environments. Our prototype, currently undergoing drone-based simulation testing, is architected for cross-platform integration with diverse aerial vehicles. CNN implementation demonstrates robust performance in adverse conditions, maintaining high classification accuracy, despite environmental variability. System architecture incorporates a comprehensive interface facilitating image corpus management, neural network training parameters, and deployment configurations across operational scenarios. Performance metrics indicate successful subject identification and tracking capabilities within specified computational constraints. This work contributes to the emerging field of mobile biometric surveillance by addressing the unique challenges presented by aerial perspectives and movement- induced image degradation, while maintaining processing efficiency suitable for embedded computing environments.
Keywords: airborne system, computer vision interface, face recognition, machine learning, real-time tracking
Autonomous Control and Navigation System: Autopilot for Drone
ME-A-04
Aviel Kriev; krievav@ac.sce.ac.il Dor Feroza; dorfe@ac.sce.ac.il
Advisors: Dr. Shayke Bilu1, Mr. Danniel Stoller1 1SCE - Shamoon College of Engineering, Ashdod
This project presents the development of a sophisticated, autonomous drone control and navigation system. The implemented architecture facilitates independent waypoint navigation, while simultaneously performing real-time obstacle detection and avoidance. It's primary innovation lies in its multisensory integration framework, which combines an advanced flight controller, GPS modules, and a camera. This integration enables awareness, obstacle identification, and dynamic flight path recalculation. Our solution surpasses conventional satellite navigation limitations by incorporating real-time image processing algorithms, enabling intelligent trajectory planning. Experimental validation demonstrates enhanced, independent operational capabilities in challenging terrain. Technical emphasis was placed on sensor fusion methodologies, computational efficiency in data, and navigation precision metrics. This implementation leverages vision techniques for object recognition, significantly advancing autonomous drone navigation capabilities in non-ideal conditions.
Keywords: autonomous drone, obstacle avoidance, real-time image processing, sensor fusion, waypoint navigation