Efficient Object Detection for Micro Aerial Vehicle Navigation Using Enhanced SSD-HOG Descriptors

Abstract

Abstract Autonomous robots, such as micro aerial vehicles (MAVs), require object detection for navigation and inspection tasks. However, the limited computational resources and real-time constraints of MAVs make object detection challenging. To address this, we propose an efficient object detection method for MAVs using enhanced SSD-HOG descriptors. Our method combines HOG and SSD techniques to create enhanced descriptors that provide better object detection accuracy and efficiency than traditional HOG or SSD descriptors. We evaluate our method on an aerial image dataset and compare it with state-of-the-art methods like SSD and YOLO. Our experimental results demonstrate that our method achieves high accuracy and real-time performance while using limited computational resources. Our proposed method is ideal for MAV navigation applications that require real-time object detection with limited computational resources.