基于SSD的轻量化无人机视觉定位算法

Vision localization algorithm of lightweight UAV based on SSD

随着深度学习的广泛应用,目标检测算法愈发丰富,相比于传统视觉识别特征方法,深度学习算法具有更高的鲁棒性和精确度。当前无人机定位算法多采用激光雷达SLAM、传感器检测等方式构建三维信息,较容易受到外界环境变化干扰。本文提出了一种基于深度学习目标检测的视觉定位算法,在外界环境受到干扰时具有较强的环境适应性,并可通过赋予图片一定含义而具有可拓展性。

With the wide application of deep learning,there are more and more object detection algorithms.Compared with traditional visual recognition feature methods,deep learning algorithm has higher robustness and accuracy.At present,the localization algorithms of unmanned aerial vehicle (UAV) are mostly based on SLAM and sensor detection,which are easily disturbed by the change of environment.In this paper,a vision localization algorithm based on deep learning object detection is proposed,which has strong adaptability to the environment when the environment is disturbed,and can be extended by giving images a certain meaning.