A Video Parallel Retrieval Method Based on Deep Hash

This paper designs a parallel video retrieval based on Spark and deep hash. The method comprises deep feature extraction using a convolution neural network based on partial semantic weighted aggregation; filtering features of image information in deep networks; the extraction and distributed storage of video summary keys; the establishment of distributed product quantitative hash coding model of image, realizing the distributed coding compression of high-dimensional features. The video parallel retrieval method proposed in this design has the advantages of high retrieval accuracy and good retrieval efficiency.