基于深度学习的智能无人果蔬售卖系统

An Intelligent unmanned fruit and vegetable vending system based on deep learning

摘要:

随着科技的不断进步,无人零售成为了未来社会的发展趋势。然而传统的无人售货超市存在成本高昂和水果蔬菜等生鲜产品售卖方式单一，效率低下，需要大量劳动力等缺陷。为了解决次问题，本文提出一种基于深度学习的智能无人果蔬售卖系统。该系统能够识别未打电子标签的果蔬生鲜，基于改进的目标检测算法精准识别果蔬的类别和位置，进而进行称重并计算商品的价格。通过实验验证，该系统具有综合化、智能化和专业化的优点，在自制果蔬数据集上，算法平均精度(mAP)提升了7.37%，实验证明本系统对果蔬的目标检测具有较强的鲁棒性。

Abstract:

With the continuous progress of science and technology, unmanned retail has become one of people's prospects for the future society. However, traditional unmanned supermarket has the defects of high cost, single selling method of fresh products such as fruits and vegetables, low efficiency, and large labor force. In order to solve these problems, an intelligent unmanned fruit and vegetable selling system based on deep learning emerged. The system can identify the fresh fruits and vegetables without electronic labels. Based on the improved target detection algorithm to accurately identify the category and location of fruits and vegetables, and then weigh and calculate the price of the goods. Through experimental verification, the system has the advantages of synthesis, intelligence and specialization. The average accuracy (mAP) of the algorithm is improved by 7.37% on the self-made fruit and vegetable data set, and the experiment proves that this system has strong robustness for target detection of fruits and vegetables.