Research And Implementation Of Image Retrieval Technology For Army Equipment Based On Deep Learning

In order to meet the needs of military reconnaissance and intelligence processing tasks under the new situation,it is a crucial step to combine emerging technologies such as big data and artificial intelligence with military theoretical requirements and flexibly apply them into the intelligence system.Deep-level neural network can dig out potential hidden information from large-scale intelligence data sets such as text and images,which contribute to the military revitalization strategy through science and technology.This paper takes image data as a subject matter of research to learn the internal rules of image data through training deep network and implement an image retrieval system,which provide a basis for strategic decision-making.This paper analyzes the problems and limitations of traditional image retrieval algorithms,and an image retrieval algorithm based on residual attention network and depth hash algorithm is proposed by analyzing the image retrieval problem of army equipment.Firstly,this method extracts image features based on the residual attention network to obtain the high-level semantic feature data of the image.Then,the features of the image are mapped into compact binary hash codes through the deep hash algorithm based on convolutional neural network.And Then,the Hamming distance formula is used to estimate the candidate result set that is similar to the army equipment image to be queried and the first N images that are most similar are found.Finally,we can use the Euclidean distance formula to reorder and sort by size to get the final result.The end results of the test on the army equipment image data set indicate that the way in this paper can combine the advantages of deep network model and deep hash algorithm.Compared with the residual network and other hash algorithms,the algorithm proposed in this paper has a higher accuracy and relatively stable retrieval performance.