Design and Implementation of an Object Detection and Localization system based on Camera-IMU and sensors
Loading...
Date
2017-06-10
Journal Title
Journal ISSN
Volume Title
Publisher
UNIVERSITY of M'SILA
Abstract
Computer vision is an important research field, widely used in a multiple of applications, from automatic image labelling to robotics. Robot and vehicles are increasingly being automated, and the need to recognize objects in the environment is essential for their correct functioning, and with the advancement in System-on-Chip technology, shrinking the size, increasing the performance and improving power consumption, allowing for whole self-sustaining intelligent systems on small mobile platforms. Another use of visual data is visual odometry which deals with movement estimation, it is helpful in GPS-denied environment, since IMU data diverges in time so VO is used to stabilize the values. In our work, we implemented an object detection and localization system on an embedded board. We explored training our own classifier for object detection, and implemented and did some testing on a VO method and collected some interesting results.
Description
Keywords
Computer vision, Object detection, Visual odometry, Raspberry Pi, Python