基于深度学习融合全景拍摄技术进行车辆识别

Vehicle Recognition Based on Deep Learning Fusion Panoramic Shooting Technology

现如今,VR技术迅速发展,社会各方面对于VR技术的实践层出不穷,本论文意从拍摄地全景视频出发,再加上较为成熟的识别和追踪算法,对于车辆进行一个便利和精准的追踪。全景视频被广泛地应用在交通和安全领域,通过实现摄像头的移动使被拍摄物体(车辆)始终居于画面中央,减少了不必要的干扰,可以尽量保证主体的完整,通过全景视频可以使相关的数据分析更加便利和精准。本系统在全景视频中检查运动目标的轨迹,运用深度学习的框架来提高车辆追踪的精度和速度。

Nowadays, with the rapid development of VR technology, the practice of VR technology in all aspects of society is endless, this paper intends to start from the panoramic video of the shooting location, coupled with a more mature identification and tracking algorithm, for a convenient and accurate tracking of vehicles. Panoramic video is widely used in the field of traffic and safety, by realizing the movement of the camera so that the object (vehicle) to be photographed is always in the center of the picture, reducing unnecessary interference, can try to ensure the integrity of the subject, through panoramic video can make the relevant data analysis more convenient and accurate. This system examines the trajectory of moving targets in a panoramic video, and uses a deep learning framework to improve the accuracy and speed of vehicle tracking.