Research And Design Of Instrument Detection System Based On Image

In a complex industrial environment,traditional instrument detection is mainly manual.Although this method is accurate,it is labor intensive and low in efficiency.In order to solve the problem of dashboard detection,in the positioning algorithm,the deep learning algorithm based on the single-shot Multi Box detector(SSD)is optimized to improve the positioning accuracy of the dashboard.In the reading algorithm,optimize the parallel refinement algorithm to eliminate distortion and reversely expand the reading accuracy.Finally,a dashboard detection system based on integrated learning ideas is proposed.1.Aiming at the complex problems of industrial environment,a target detection algorithm based on deep learning is proposed to train images in extreme environments to solve the problem of instrument positioning.First,the SSD basic network is replaced with Res Net-50,which reduces the computational complexity while enhancing the network feature extraction capabilities;then adds a feature pyramid network structure,which combines the detailed information at the bottom of the network with the semantic information at the high level to enhance the detection of small targets;Finally,GIo U is introduced in the position regression as a new metric to compare the two shapes,and the conventional Smooth L1 loss is replaced with GIo U loss to accelerate the convergence of the model.Experimental results show that the model has high accuracy for instrument positioning.2.Aiming at the problem of distortion and non-single pixel in Zhang’s rapid parallel refinement,the paper proposes an improved algorithm.The corresponding operation is made by judging the stroke of the pixel to improve the distortion problem,and then the second refinement is used to solve the incomplete refinement of the original algorithm.The experimental results show that the refined curve is guaranteed to be single pixel and no distortion,which effectively solves the problem of pixel redundancy.Finally,the improved algorithm based on Zhang’s rapid parallel refinement is applied to the reading recognition of analog industrial instruments.The experimental results show that the accuracy of the meter reading is improved.3.The algorithm proposed in the paper is applied to the actual industrial environment,and a real-time monitoring system for industrial instruments is designed and implemented.Use Py Qt5 and Py Charm to develop the front-end interface,transplant the algorithm to the server,and use Socket for communication.The server collects instrument images through the remote monitoring system and stores them in the database,and the client can call the integrated algorithm to read and recognize the collected images.Finally,the system function test is carried out to test the feasibility and effectiveness of the system.