Smoking behavior real-time detection method based on deep learning

Abstract

The invention is suitable for the technical field of artificial intelligence and behavior detection, and provides a smoking behavior real-time detection method based on deep learning. According to themethod, a data set is constructed through a data acquisition standard, and a set of smoking convolutional neural network for smoking behavior detection is constructed. The method promotes research ofdeep learning on smoking behavior detection, and has good detection accuracy, and by means of deep learning, the method has very high real-time performance and robustness and can adapt to the environmental change of an application scene.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence and behavior detection, and in particular relates to a real-time detection method of smoking behavior based on deep learning. Background technique [0002] Human action recognition is a fundamental task in computer vision. Vision-based human action recognition is the process of labeling image sequences with action labels. Reliable solutions to this problem have been used in video surveillance, video retrieval, human-computer interaction, and video understanding, etc. It has been applied in the field, for example, video surveillance based on behavior recognition can be applied to the smart home of the elderly, and the elderly fall and other problems. Video retrieves a tackle in a football game, a handshake in a news footage, or a typical dance move in a music video. Interactive applications, such as those in human-computer interaction or games. This task is very challenging due to...

AI Technical Summary

Problems solved by technology

The algorithm based on this method has a single feature extraction, so it generally has poor generalization ability, and the accuracy of smoking behavior recognition in complex environments is not high; the establishment of the model is complicated, and it does not have high real-time performance in actual application. sex

The other is to extract features from video stream images with the help of convolutional neural network, locate the smoking position in the image, and judge the behavior category. This technology can extract rich smoking behavior features with relatively good accuracy, and the deep learning algorithm It generally has high real-time performance, but due to the lack of relevant data sets, the research is not deep enough, and the trained model does not have good practical application capabilities; and the application of convolutional neural networks is only to apply the existing network, and there is no Construct a dedicated network for smoking behavior detection based on smoking characteristics

Images