Sihai network

Where is Hangzhou unmanned supermarket? How does the unmanned supermarket charge? Is there really no

Alibaba's unmanned supermarket recently appeared in Hangzhou. Where is the unmanned supermarket? What is the black technology inventory of Alibaba's unmanned supermarket? Many people are most tired of going to the supermarket and many waiters are chasing you to sell things, but Alibaba's unmanned supermarket avoids this. Do you know the biggest highlight of the unmanned supermarket? How to calculate the charge when no one supermarket goes out? Follow Xiaobian to learn about this new thing.

Just over a month later, Alibaba's own unmanned retail store "taocoffee" was ready to come out. It is not only imaginative, but also whether Alibaba's unmanned supermarket project cooperates with Shenlan technology? Lei feng.com asked Chen Haibo, founder of Shenlan technology, about this, and he said that the two companies did cooperate on this project. He also revealed that there are some similarities between the scheme used by "coffee shopping" and the scheme used by quixmart. (this article is exclusively compiled by WYH, editor of sihai.com. If you need to reprint it, please indicate the source and source. Related article link: http://www.4hw.com.cn/life/zone.html)

Mr. Chen Haibo introduced in an exclusive interview with Lei Feng. Com that quixmart used convolutional neural network and machine vision to identify and analyze consumers' bodies and commodities. He said that because the face is easy to be occluded, body recognition is more feasible than face recognition in the continuous tracking of consumers, which is not easy to lose.

He also said that the biggest challenge for machine vision to identify goods is to remove the background. The change of light in the store and the mobile consumers will affect the accuracy of recognition.

The head of Alibaba's unmanned retail team, DOMA, also mentioned this in an interview. He said that at present, Alibaba adopts a combination of multi-channel cameras and sensors. The main function of multi-channel cameras is to monitor the situation in the store and track and analyze the behavior of consumers. For example, by analyzing the expression and body language of users when they get a certain product, we can help businesses judge whether a certain product is enough to meet their wishes; by capturing the movement track of consumers in the store and the length of stay in front of the shelf, we can guide businesses to adjust the display mode and service devices in the store. But unlike quixmart, the purchase of goods in Alibaba unmanned supermarket is not completed through machine vision system. Although the identification system is also integrated on the shelf, it is mainly used to record the behavior of consumers, such as how many times consumers pick up and put down a specific product. DOMA said that there is a loophole in the current shopping on the shelves.

According to the insiders, with the increase of store area and traffic, the unmanned retail stores adopting machine vision will pose a huge challenge to GPU, and the recognition accuracy is difficult to guarantee. (this article is exclusively compiled by WYH, editor of sihai.com. If you need to reprint it, please indicate the source and source. Related article link: http://www.4hw.com.cn/life/zone.html)

Mr. Chen Haibo introduced that in order to ensure the stability of the system of quixmart's self-service convenience store, it can also work normally in extreme weather and poor network environment, and its identification and calculation work are carried out locally. According to Chen Weilong, CEO of tianruo technology, localization may be necessary, because it is impossible to control where such space is placed. The state of the network is likely to be unreliable, and the amount of calculation is so large. If it is placed in the cloud, it will be impossible to use it.

Chen Weilong also said that the accuracy rate of the image recognition system used in the convenience store is acceptable in a small range of scenes -- such as 10 square meters, 100 commodity categories, 2 users -- and high capital and technology investment. But if we want to expand further, we will face great challenges. Considering that the area of Alibaba's unmanned retail store is as high as 200 square meters, it is difficult to use machine vision to purchase goods, either from the challenge to GPU or from the accuracy of identification. (this article is exclusively compiled by WYH, editor of sihai.com. If you need to reprint it, please indicate the source and source. Related article link: http://www.4hw.com.cn/life/zone.html)

So how can we confirm the goods that consumers want to buy? When consumers leave the store, they need to go through a 'settlement gate' composed of two gates. When the first door senses the user's departure needs, it will automatically open. After a few seconds, the second doors will be opened. At this point, the charge has been completed. The prompting device on one side will tell the user, 'Alipay total deduct XX Yuan'. Commodity identification is completed between two doors. It is not known whether RFID or machine vision technology is used. However, Ali may also have a big idea, because DOMA said in the interview that the commodity identification technology is still sensitive, involving patent applications.

Alibaba's settlement plan is also different from that of quixmart. When quixmart enters the store and settles accounts, it uses capillary recognition in the palm of the hand, while Ali unmanned supermarket uses "non cooperative face recognition" technology, that is, the system can accurately capture the biometrics of the user without looking at the lens, and carry out matching and calculation. When users enter the store for the first time, they need to collect face information and agree to the automatic deduction protocol. However, there is a ceiling on the amount of withholding. The specific ceiling varies according to different scenarios. It is reported that the ceiling of withholding on the creation Festival is 5000 yuan.

At present, one of the common problems of the unmanned retail stores is that they support a small number of commodity categories, and Alibaba's "coffee shopping" unmanned supermarket is no exception. According to DOMA, the products in the "coffee shopping" unmanned supermarket are all limited products on the creation Festival. But according to Chen Weilong, this kind of convenience store needs to reach at least 50 square meters, 300 commodity categories (and can't be specific), so it has application value. (this article is exclusively compiled by WYH, editor of sihai.com. If you need to reprint it, please indicate the source and source. Related article link: http://www.4hw.com.cn/life/zone.html)

Although taocoffee is known as 200 square meters, a large part of the 200 square meters may be the rest and catering area, and there is not a lot of space for shelves.

From all kinds of signs, Alibaba unmanned supermarket is still just a testing ground of new technology, mainly used for data collection and analysis, which is still a long time away from commercial use. After all, so far, Alibaba's face payment has not been officially commercial.

DOMA said frankly that in the part of commodity identification, what Ali can do is that the user does not need to make the commodity deliberately face the device, but also can be identified, but the premise is that the consumer should cooperate, and do not deliberately make the commodity avoid being identified. In this way, how to supervise consumers has become a major issue, so Alibaba has also adopted some mechanisms. When settling accounts, if consumers try to cheat, such as deliberately hiding goods or tearing up electronic identification codes, it will be regarded as a bad record and affect their next entry.

When asked about the specific business hours of the store, DOMA said that the store solution does not mean that there is no one at all, but just that there is no cashier. There can still be shopping guides in the store, just to use new technology for better service, which is Alibaba's current vision.

So I'm looking forward to Ali's showing a lot of "black technology" and bringing the unmanned supermarket to your door. This time I may be disappointed.