基于Hadoop的分布式视频识别框架

With the increasing demand of video recognition, a highly efficient and reliable framework for large-scale video recognition is needed to facilitate the development of related applications. The existing Hadoop distributed framework has been shown to have good stability and outstanding performance in the application of read and write intensive data processing. But for the computation intensive video recognition applications, no relevant framework can be directly utilized. Based on the Hadoop framework, and combining with the feature of video recognition task, a distributed video recognition framework is proposed, which includes video storage and video recognition. The experimental results show that the proposed framework has a good performance in the storage stability and the scalability of computation.