A gesture recognition method and system based on stt-lstm network

Abstract

The invention discloses a gesture recognition method based on STT-LSTM network, and constructs an STT-LSTM network model, which contains two layers of STT-LSTM networks; RGB features and optical flow features are respectively transmitted into the first layer of STT-LSTM network , in the training process of the first layer of STT‑LSTM network, initialize the global context information, and pass the initialized global context information to the second layer of STT‑LSTM network, and continuously carry out in the second layer of STT‑LSTM network Iteratively update, gradually improve the context information, reorganize the joint points finally output by the second-layer STT‑LSTM network, form a node pipeline containing gesture action information, and use it as a classification label for multi-classifiers to predict traffic police gestures. The method of the invention can accurately identify gesture command actions, can effectively ensure the smoothness and safety of traffic, and reduce the probability of accidents.

Description

technical field [0001] The invention relates to the technical field of image and video analysis, in particular to a gesture recognition method and system based on an STT-LSTM network. Background technique [0002] With the rapid development of my country's automobile industry, traffic congestion and traffic safety problems also follow. The importance of traffic police in the traffic industry is also growing. Its main responsibilities are to maintain traffic order, deal with traffic accidents, check and correct road traffic violations, and be responsible for the registration and management of motor vehicles. The traffic police gesture is an important tool for the traffic police to ensure the smooth and safe traffic. [0003] The following problems are often encountered when driving a motor vehicle: First, nearly 60% of drivers cannot fully recognize the meaning of the traffic police's command gestures, which reduces the efficiency of the traffic police's command and cannot g...

AI Technical Summary

Problems solved by technology

[0003] The following problems are often encountered when driving a motor vehicle: first, nearly 60% of drivers cannot fully recognize the meaning of the command gestures of the traffic police, resulting in a reduction in the command efficiency of the traffic police, and driving safety cannot be guaranteed; secondly, when driving in rainy and snowy weather or at night, The visibility is low, and the gestures of the traffic police are difficult to recognize; moreover, when the traffic flow is heavy on the traffic control section, due to the blocking of vehicles, the command distance of the traffic police gestures is limited

Images