Design And Implementation Of Visual Recognition System For Safety Elements In Digital Workshop

Recently,the domestic manufacturing industry began to develop towards networking and digitization.As the core unit of intelligent manufacturing,digital workshop begins the development of digital innovation.Safety in production is the prerequisite for development,and the workshop safety can not be ignored.It is necessary to give consideration to both development and safety at all times.In this thesis,Fully Convolution One-Stage object detection algorithm is used to detect and recognize the safety elements in the digital workshop.The safety elements are divided into the unsafe behavior of people and the unsafe state of objects.The feasibility of the algorithm is verified by selecting an example from two aspects,that is,the detection and recognition of the digital reading of industrial instruments in the digital workshop,and the safety helmet wearing of the personnel in the digital workshop detection and recognition,and different solutions are formulated according to the respective characteristics of the two scenes.In the application scenario of industrial instrument reading detection and recognition in digital workshop,because the target occupies a small area in the detection range,Fully Convolution One-Stage object detection algorithm is used to locate and segment the digital display area of industrial instruments to narrow the recognition range of instrument readings.Then the full convolution neural network is used to recognize the digital display readings of industrial instruments.Finally,the reliability of the algorithm is proved by experiments.In the application scenario of the detection and recognition of the safety helmet wearing of the personnel in the digital workshop,Fully Convolution One-Stage detection algorithm is used to detect whether people in the digital workshop wear safety helmet.Because the background environment is complex and the shape of the helmet is changeable,after the original feature extraction network structure of the algorithm,the balanced feature pyramid is added to make full use of the extracted features of different levels,and the Gaussian non local attention mechanism is added to further refine the integrated features to improve the performance of the model.Finally,the reliability of the algorithm is verified by experiments,and the target detection evaluation index is given to evaluate the performance of the model.Finally,the visual recognition system of digital workshop safety elements is designed and implemented.After the demand analysis and module design of the system,the system is divided into five modules,namely camera module,algorithm module,database module,user interface module and alarm module.The functions of each module are introduced in detail,and the usability of the system is verified by system test.