Research On Attention-Based Video Summarization And Pedestrian Attribute Recognition

In real life,human brain can use the limited resources to quickly screen out the most valuable information.This means is called attention mechanism.Recently,the attention mechanism has been widely used in the field of deep learning such as natural language processing,speech recognition,and image recognition.Attention mechanism can improve the performance of a model by making more efficient use of its input and output.This paper first introduces attention mechanism into the field of video summarization,which proves the feasibility of attention mechanism in the video summarization task.Further,a pedestrian attribute identification method based on attention mechanism that maximizes the use of input and output information is studied.With the increasing number of videos,how to effectively store and process video data has become an urgent problem to be solved.Video summarization technology extracts information-rich video clips or video frame sets from the original video,which can quickly summarize important video information,thus alleviating the problems caused by the explosion of video.Aiming at the problem that existing video summarization algorithm ignore the semantic association between video frames,this paper proposes a video summarization algorithm.The convolutional neural network is used to encode the original video,and the attention-based neural network is used for decoding.An attention mechanism is used to fully explore the interrelationship between video frames to generate an information-rich summary.The validity of the proposed model is verified on the manually labeled SumMe and TVSUM datasets.The surveillance video network of city provides an important guarantee for maintaining urban security.Pedestrians are the main target of monitoring systems.It is of great practical significance to be able to automatically identify the individual attributes of pedestrians.Most of the existing pedestrian attribute recognition methods transform the pedestrian attribute recognition problem into a multi-classification problem,ignoring the relationship between attributes,which is not conducive to the accurate identification of attributes.This paper proposes a pedestrian attribute recognition algorithm based on joint guidance attention mechanism.The algorithom using visual and attribute features to guide the attention-based decoder to generate attributes,which makes full use of the connection between vision and attributes,attributes and attributes to achieve the purpose of accurately identifying pedestrian attributes.A lot of experiments have been carried out on the existing two pedestrian identification data sets PETA and RAP,and the effectiveness of the proposed algorithm is verified.