Xidian undergraduates develop fruit and vegetable self-service electronic scales, use image recognition technology to weigh and settle

Project Introduction: This project is a self-service electronic scale that can simultaneously identify and weigh different fruits and vegetables. The key technologies are multi-target image recognition and distributed weighing. Recognize and divide the contours by reading the image information of fruits and vegetables during work; combine the feedback of multiple weighing modules to determine the quality of each fruit and vegetable at different positions, and finally display the consumption information on the terminal and generate a payment QR code, the customer scans Payment can be completed. This project fills the gap of the intelligent fruit and vegetable weighing system on the market, superior to the existing single intelligent electronic scale, and highly competitive in the market.

The core department of the team: School of Electronic Engineering, Xidian University

In recent years, the term "unmanned supermarket" has frequently appeared in people's eyes. In the streets of Hangzhou in 2017, Ma Yun ’s first unmanned supermarket officially opened, which opened the curtain for the development of unmanned supermarkets-no cashier, no checkout, no queuing, 24-hour sales ... countless subversive words are full of People's eyeballs. Over time, unmanned supermarkets may really replace traditional supermarkets. Replacing salespersons with intelligent electronic scales to do a single repetitive job is the most important part of an unmanned supermarket. According to this, Liu Zhaoxin and his "Gravity" team combined with the most popular artificial intelligence to bring a multi-functional self-service fruit and vegetable settlement system that integrates weighing, identification, measurement, and settlement.

On December 13, 2017, at the site of the public defense of the 29th Spark Cup, Xi’an University of Electronic Technology, the self-service fruit and vegetable settlement system brought by Liu Zhaoxin and his team, relying on solid technology, rigorous attitude and various commercialized The consideration won the unanimous praise of the judges present, and finally won the good results of the special prize.

From love, loyal to technology

Liu Zhaoxin, as a student majoring in electronic information and majoring in electromagnetic field microwave, how did he think of using algorithm recognition to build such a system? Creation comes from life. When the initial team had only three members, one of the parents did the weighing work in the supermarket. She inadvertently brought the trouble of weighing work to her children. The idea of ​​using existing technology to make an intelligent electronic scale to replace human work has arisen. As the captain of the "Gravity" team and an algorithm designer in the entire project, Liu Zhaoxin greatly exerted his love and attention for artificial intelligence. Based on his interest and ability, he led the team members to carry out exploration and research.

Dare to dare to do it, just do it, and they start with enthusiasm. With the tacit cooperation of the team members, the first generation of works can be completed easily. However, the success rate of image recognition can only reach 67%, which is not consistent with the expectations of the team members and the acceptable range of the market. If a person wants to climb higher, it means he has to sweat more, do more work, and work harder than ordinary people. The "Gravity" team explains this well. In the constant communication with the team instructor Professor Shi Guangming, they found that the CV algorithm they used did not solve the image recognition problem well. Now the industry generally uses deep learning to solve this problem. The team finally decided to use the most popular deep learning technology in artificial intelligence to upgrade the first generation of works. But the process of transformation is not smooth, and because of the lack of theoretical basis and practical experience, it is difficult for them to control this cutting-edge theory. Teacher Shi carefully guided them to learn theoretical knowledge, and provided important help in key links such as technical solutions and product planning. After replacing the traditional CV algorithm, the success rate of the system in the field of image recognition has been increased to 99.95%. So with the new generation of self-service fruit and vegetable settlement system of "gravitation", an artificial intelligence product that has been explored extremely profoundly in terms of hardware structure and business context, and reached a new level of stability.

Faith will move mountains to open

Recalling this road, the difficulty was like a hill on the road of "gravity"-because the computer system does not support the algorithm version, Liu Zhaoxin sacrificed his winter break to go to the high school counseling class to modify the code, "At that time It ’s almost the New Year, and their high school students are gone. I ’m still writing code there. It ’s still pretty miserable. Even if you hear the sound of firecrackers outside, you still have to write the code. In the end, the air conditioner is gone and you have to write the code The harvest was pretty big at that time, "Liu Zhaoxin said. Faced with difficulties, they chose to face the difficulties and never thought of giving up, just because they liked it, so the difficulties became a rare opportunity for exchange and learning. "It's actually an honour to have such an opportunity. It's an honour to be able to exchange learning algorithms with the big brothers, and Mr. Shi often has some constructive opinions."

From a simple student work to a product with a simple business model, operational thinking, and unique market role, the self-service fruit and vegetable settlement system has experienced great breakthroughs in technology through the mass production of circuits and chips. Therefore, the entire stability and accuracy have been qualitatively changed.

The stars in the sky are also dotted with night, so why only one of them? According to Liu Zhaoxin's own words, on the one hand, because their work focuses on application scenarios compared to others; on the other hand, because the work itself is extremely stable, there are no errors in the external display.

Looking back at Zhengrong, the heart is endless

Talking about the first prize of the 28th Spark Cup, Liu Zhaoxin was a little more relieved: "After the first idea, everyone just made their own, a little slack. Until the night before the Spark Cup handed over the work We haven't finished it yet, but when we got together, we actually finished the work overnight. I slept for two hours, the member Cheng Yuheng who was in charge of the assembly did not sleep overnight, and then went to the defense the next day. I thought the teacher might be That is to encourage us and give us a first prize at the school level. "But that was the first hasty work, very immature. After a year of re-exploration, they created a more high-end work. Today's outstanding results.

Such excellent commodities are also faced with many problems in the process of realizing commercialization in the market. From the product itself, the mass production requires the re-optimization of the algorithm, hardware or mechanical structure of the product in order to pursue higher Of stability.

The expectation of "Gravity" is that in the far future, this self-service fruit settlement system will no longer be limited to fresh food, but will open up a broader space for development in bulk snacks. At present, they are negotiating with PNP and other investors, and are also making further technical barriers, and will expand their investments when the technical products are stable to a certain degree.

Liu Zhaoxin quoted Ma Yun to encourage himself-subversive things are what he really wants to do. I believe that in the near future, our shopping habits will also change due to it.

The golden needle is the person who asks for the sky

Along the way, they did not exquisite and simple patchwork when they entered the freshman year, to high accuracy, to the integration of commercial elements and works, and finally formed product thinking. All this is due to their own efforts, teacher guidance and team cooperation.

Asked about the suggestions for inexperienced contestants, Liu Zhaoxin emphasized that the practical value of the created works should be found, the thinking should be active, and the creative process should be taken seriously. "Working on a project is almost the same as learning. You need to concentrate on doing it yourself, accumulate slowly, grow from failure, and eventually achieve the effect of improving your own ability."