Research And Application Of Image Object Detection And Segmenttation Algorithm Based On Deep Learing

Image object detection and segmentation plays a key role in the field of image processing and machine vision.Because of the complexity and diversity of image information,the traditional method is difficult to achieve the ideal effect when the features of the image object are obscure.In recent years,the rapid development of deep learning has gone beyond the characteristics of artificial design in quantity and performance,and surpassed the traditional visual algorithm in the field of object detection and segmentation through the characteristics of autonomous learning under large data.In this paper,deep learning based image object detection and segmentation methods are deeply studied.At first,Grabcut algorithm is a widely used interactive segmentation method.In this paper,this algorithm needs to manually annotate the object rectangle frame and have two long running time.This paper presents an improved Grabcut segmentation algorithm based on depth learning,using the method of object detection based on depth learning and combining with the super pixel segmentation algorithm.Method automatically extracts the target box instead of the manual annotation target box.At the same time,the super pixel optimization algorithm is used to reduce the number of iterations and improve the efficiency of the algorithm.The automatic high precision segmentation algorithm based on the target detection model is realized,which makes the contour as close to the object as possible.Secondly,in view of the problem of intelligent monitoring in industrial production and the need for manual reading for small electronic devices,this paper uses the transfer learning to apply the object detection technology based on deep learning to the industrial image.The data set is created from a large number of image information,and the object detection model is generated by training optimization parameters in the tens of thousands of iterations.The object location and category information in the image are segmented to realize the intelligent monitoring of the pipeline operation.The CNN deep convolution neural network model is designed according to the actual problems of electronic equipment readings.The digital information in the standard area is accurately identified.Finally,in this paper,a large number of analysis and test experiments are carried out on the above research content.The test pictures are selected from the public data set PASCAL VOC2007,and the other part is the industrial image taken by the handheld camera.By verifying and comparing the results of this algorithm with the existing segmentation methods,it is applied to the intelligent monitoring system of pipeline operation on the premise of ensuring the correctness and efficiency of the method,which lays the foundation for the realization of intelligent manufacturing.