基于图像识别的果蔬自助结算系统

Fruit and Vegetable Self-service Settlement System Based on Image Recognition

摘要:

实现对缺少条形码的水果蔬菜的识别与结算，是超市自助结算的一大难题。为在资源有限的结算终端设备上实现超市果蔬的识别与分类，提出了一种基于神经网络的果蔬识别算法。通过增加网络宽度的方法改进Alex Net，提升识别性能。结合压力传感器、摄像头等硬件设备，在树莓派上进行实验，完成了果蔬自助结算系统的搭建。经实验测试，系统对果蔬的平均识别准确率可达98.25%，单次结算总耗时约7.48秒，仅为人工结算耗时的1/4，满足果蔬自助结算系统的实际应用需求。

Abstract:

Realizing the identification and settlement of fruits and vegetables that lack barcodes is a major problem in supermarket self-service settlement. In order to realize the identification and classification of fruits and vegetables in supermarkets on the settlement terminal equipment with limited resources, a fruit and vegetable identification algorithm based on neural network is proposed. Improve Alex Net by increasing the network width to improve recognition performance. Combined with hardware devices such as pressure sensors and cameras, experiments were conducted on the Raspberry Pi to complete the construction of a fruit and vegetable self-service settlement system. After experimental tests, the average recognition accuracy of the system for fruits and vegetables can reach 98.25%, and the total time for a single settlement is about 7.48 seconds, which is only 1/4 of the time for manual settlement, which meets the actual application requirements of the fruit and vegetable self-service settlement system.