基于深度学习的海量航拍视频智能处理技术

Intelligent processing technology of massive aerial video based on deep learning

针对现阶段处理航拍视频数据工作量大且繁琐、人工成本高、重点目标信息提取效率低以及难以有效管理等问题,文中提出一种基于深度学习的海量航拍视频智能处理技术。该技术以深度学习算法为基础,采用可变卷积网络完成视频目标智能检测识别,采用自组织特征映射神经网络完成视频片段智能切分,使用局部敏感哈希算法完成海量视频快速检索。通过视频智能处理技术能够剔除海量航拍视频中的无效视频片段,保存有效目标视频片段,提高检索效率,降低存储压力,实现对海量航拍视频数据的高效管理。在实际项目中,基于海量航拍视频智能处理技术完成航拍视频智能化处理软件开发,工程实践应用证明,该技术可以有效减少作业人员重复性工作,提高视频处理作业效率。

In allusion to the problems of processing aerial video data at this stage,such as heavy workload,high labor cost, low efficiency of key target information extraction and difficulty in effective management, an intelligent processing technology for massive aerial video based on deep learning is proposed. Based on the deep learning algorithm, in this technology,the variable convolution network is used to complete the intelligent detection and recognition of video targets,the self-organizing feature mapping neural network is used to complete the intelligent segmentation of video clips,and the localitysensitive hashing algorithm is used to complete the rapid retrieval of massive videos. The video intelligent processing technology can be used to eliminate invalid video clips in massive aerial videos,save valid target video clips,improve retrieval efficiency,and reduce storage pressure,so as to realize efficient management of massive aerial video data. In the actual project,the development of aerial video intelligent processing software is completed based on this technology. The engineering practice application proves that this technology can effectively reduce the repetitive work of operators and improve the efficiency of video processing.