Design And Implementation Of Video Semantic Analysis System Based On CNN And LSTM

At present,convolutional neural networks are a hot topic in the computer field.At the same time,it has achieved remarkable results in many tasks in the computer field.With the advent of the information age,the number of video data presents an explosive blowout.Due to the fact that people can't quickly retrieve information from video,how to apply convolutional neural networks to the tasks of video analysis,image-based recognition technology is very important.Currently,deep-learning based convolutional neural network algorithms are used in images.The recognition field has achieved good results,but there are still many problems to be solved.In order to solve the problem of convolutional neural network feature extraction,this paper improves the convolutional neural network feature extraction method,proposes a local LBCNNbased video feature extraction optimization model,effectively solves the challenges of object rotation,and the network model parameters Reducing the place where some hardware restrictions are relatively large can effectively conduct network learning and reasoning.For the LSTM network semantic recognition problem,this paper considers adding the attention mechanism to the semantic analysis problem based on video content,that is,after extracting the video image features,the video image features and the previously predicted word information are collectively input into the LSTM network and then calculated.Layer output,so that you can prompt according to the previously predicted word information should focus on which part of the video image,rather than aimlessly focus on the entire video image,the experimental results show that the model effectively improves the accuracy of semantic recognition.The feature extraction optimization model proposed in this paper and the LSTM-based video semantic analysis model have been successfully applied in the school-enterprise cooperation project based on big data video analysis.The future expectations of the video analysis system can be applied to the security analysis of related logistics videos,and the logistics transportation process.In the license plate recognition and other directions.