Research And Implementation Of Fruit And Vegetable Self-service Weighing System Based On Target Detection

The progress of science and technology should ultimately promote the development of society and improve the quality of people's life.At present,there are two ways of weighing and labeling fruits and vegetables: special weighing and customer semi self-service weighing.The special weighing method has high labor cost and is not suitable for multiple equipment.The semi selfservice weighing method is selected by the customers in hundreds of fruits and vegetables.Although multiple equipment can be put into the system,the single weighing time is longer.Both weighing methods will lead to congestion when the supermarket has a large population flow.Therefore,it is very necessary and valuable to design and develop a set of fruit and vegetable self-service weighing system based on target detection.Based on the advanced target detection technology,this thesis designs and develops a fruit and vegetable self-service weighing system for supermarkets and vendors,aiming to improve the efficiency of fruit and vegetable weighing and reduce the cost of fruit and vegetable weighing.The main contribution of this thesis are as follows:1.In view of the problem that the existing image data set of fruits and vegetables is not consistent with the actual use of the fruit and vegetable self-service weighing system,this thesis uses the field collection method to establish the image database which can be applied to the fruit and vegetable weighing system according to the actual use requirements of the system.In order to ensure the validity of data,the acquisition equipment and system equipment are the same in the process of acquisition,and the pictures have multiple angles and backgrounds.The data collected are amplified to make the training model generalized.2.In view of the traditional single image classification model can't solve the problem of multi pricing method and multi human factors in the actual fruit and vegetable system,this thesis uses the YOLOv5 fruit and vegetable target detection model to realize the detection of multi fruit and vegetable target,solve the problem of multi pricing method and multi human factors,and use ResNet to improve the accuracy of YOLOv5 to ensure the system's requirements for recognition accuracy.3.Design and implement the fruit and vegetable self-service weighing system based on YOLOv5 and ResNet.The demand of the actual users of the system is analyzed,and the structure and function of the fruit and vegetable self-service weighing system are designed.The system is mainly divided into two parts: one is customer-oriented function,including fruit and vegetable self-service weighing,manual selection of fruit and vegetable,fruit and vegetable order confirmation,etc.;the other is administrator oriented function,including administrator login registration,model expansion training,updating fruit and vegetable information,query fruit and vegetable order,etc.According to the system design,each function of the system is realized and tested.Finally,comparing the time efficiency and cost of the existing fruit and vegetable weighing system,it shows the advantages of the system in time efficiency and cost input.And the actual effect of the system is displayed.