Artificial intelligence and fashion combined with mushroom street to save straight men with AI

Mushroom Street has been studying this matter recently, and has set up a matching research institute, hoping to combine subjective aesthetics with data algorithms to more efficiently meet the needs of users at any time and any place.

Artificial intelligence gives you a set of the most fashionable wear this fall, do you dare to wear it?

Mushroom Street has been studying this matter recently, and has set up a matching research institute, hoping to combine subjective aesthetics with data algorithms to more efficiently meet the needs of users at any time and any place.

Programmers who need to solve their own clothes have compiled an algorithm to tell you the latest fashion trends. However, as artificial intelligence enters the fields of driverless, medical, and smart home, its integration with the fashion industry is not impossible.

The premise of artificial intelligence is a massive database, and the fashion industry naturally has this condition. There are hundreds of thousands of brands and designers in the fashion industry. Everyone's wisdom contributes to the fashion industry, and together they become a fashion trend. Wang Ying, director of fashion content at Mushroom Street, explained in detail: “The designers’ brains are all inspired by the senses, not the real sense of the sky.” The so-called “reliable inspiration”, for example, even a rich Experienced fashion magazine editors have accumulated this fashion sense through years of hard work and countless work experience. This process is similar to the deep learning of artificial intelligence.

The advantage of Mushroom Street is that the flow of e-commerce platforms makes these stylish data more aggregated and presentable. Before the landing of "AI+ wear", the main result of the mushroom street research institute was the release of the 2017 autumn and winter fashion trend report. Some key elements mentioned in the report, such as plaid, retro, hip hop, etc., the main data sources include the traditional fashion show of major fashion brands, street shooting (Mushroom Street team in Beijing, Shanghai, Guangzhou and other trend gathering places Large-scale collection of pictures and some reds, talents, and a large part of the data flow from the mushroom street online platform, including a variety of related content and social, user browsing, purchase, repurchase Data, these are traditional fashion brands, stores do not have.

Another source of data for the mushroom street is the clothing wholesale market in Guangzhou and Hangzhou. The invoicing data of those who are in the first line and face the users can also explain some trends. Overall, it can be regarded as a very grounded fashion report.

The significance of this report is not only to help the platform to better arrange and design the layout. The key point is that it is an attempt to integrate online and offline traffic data. In the current popular words, AI's scene application.

Zhang Hongming, an expert in image algorithm technology at Mushroom Street, showed us the specific ideas for future AI wear:

When you choose a top, some tangled lower body how to wear it, you can take a photo with your mobile phone. After uploading, the system will automatically identify, analyze and process all kinds of elements in this picture: first, a goddess, the top has Some basic elements: black, lace shirt (but how do you think it should be a chiffon shirt). In fact, this jacket also has an important element - printing, which has not yet been identified, which is necessary to be refined in the future.

According to these elements, the system recommended a combination of hip skirt, A-line denim skirt, wide-leg pants, and flared pants. Chiffon with tannins, a horse and a tiger to give this AI an 80 point.

Presenting such results requires three solutions: massive image data, fashion analysis algorithms, and rich application scenarios.

At present, the image data of this AI technology of Mushroom Street comes from the existing SKU on the platform. See the favorite recommendations in the future, you can also buy with one click. In the future, there will be pictures uploaded by users in the community, as well as some poster pictures to enrich the picture library. Compared to the shopping function, the application of Mushroom Street still wants to focus on the recommendations.

The fashion analysis algorithm sounds very high-end, in fact, it is a multi-dimensional analysis of the picture. Zhang Hongming takes this picture as an example:

First, extract the area: tops, pants, bags, shoes. Then extract the attributes of these four areas: including styles, colors, and elements. For example, the top is a round neck, letter T-shirt. For the human brain, what you see at a glance, the machine seems to be a little bit cute. However, in the recognition of colors, the system can accurately calculate the color index (if the problem of shooting light is excluded), which is the place where the current machine is more powerful than the human brain. The analysis of these basic information is the basis for all the recommended work.

Zhang Hongming also mentioned at the press conference: "The fashion trend is very complicated, involving changes in different seasons and different regions. Our collocation principle is to meet the e-commerce platform, the young women after the Pan 90 Specifically, the algorithm of Mushroom Street is for its own users. “We will conduct group analysis and personalized preference analysis. The core of the e-commerce platform – the fabrics and materials of the products will also be taken into consideration.”

However, the plan to wear on the AI ​​is still in the process of incubation. At present, some of the works that it has completed can only be regarded as quite satisfactory. It is not said to be very good, but it will not go wrong. At least the moderate users of the mushroom street to the heavy users (about half of them) can reach the level of artificial intelligence. Zeng Xianjie, vice president of Beauty United Group, attributed it to the general status of AI technology development: "At present, the degree of artificial intelligence development can reach better than the average person, but there is still a gap compared with experts."

If it is said that Mushroom Street wants to tell users how to wear it through this technology, it seems to be somewhat unpleasant. The age group of Mushroom Street is a younger group of women aged 16-23. They grew up in the era of information explosion. They are knowledgeable and assertive. If a jacket can only provide four pieces, this new Things are obviously going to be abandoned.

However, the core of AI technology is to save a lot of repetitive work and better meet the diverse needs of users. If the data is enough, the future AI wear technology can more accurately locate your needs: small girls, high, thin, music festival wear and so on. This is like removing a lot of invalid information in the search, which improves the efficiency of the screening.

A variety of integrated, vertical e-commerce platforms, the most talked about is the conversion of traffic. At present, the number of registered users of Mushroom Street exceeds 200 million, and the daily activity is 10 million+. The small program is online for two months. The number of newly purchased users exceeds 3 million. From mid-July to early August, users increased by 300%. Zhang Xin, director of fashion goods at Mushroom Street, believes: "The vertical platform will develop into a certain stage and will definitely penetrate into the core of the industry and understand the industry more deeply." 36æ°ª learned that for this technology, Mushroom Street does not just take it. It is regarded as a traffic introduction mechanism. In the future, these accumulated data will be applied to all aspects of design, production, and supply chain. In fact, it is equivalent to a service provided to merchants.

The main users of Mushroom Street are after the age of 16 to 23 years old. In Accenture's latest "post-95 consumer report", 31% of the 95s still prefer to go to the store to shop, which is higher than the mobile phone. However, this data does not need to make the e-commerce platform such as pin-and-needle. It is better to say that it is because today's stores can provide various digital experiences. If the AI ​​wear technology can only be browsed and recommended, it is only a very advanced search. What is important is how to serve more complex real-life scenarios.

Editor in charge: Wang Zhen