As one of the focus of retail technology in recent years, smart retail technology has attracted much attention.
In mid-January, heavyweight players in the retail industry gather at the NRF retail show. As the world’s largest retail exhibition, NRF’s Annual Convention & EXPO has successfully held 109 sessions. This conference, which aims to discover and discuss the latest technology, opportunities, and practices in the industry, is now known as the “Retail’s BIG Show” and is also considered a “treasure place” for the latest retail technology.
This year, NRF 2020 not only welcomed more than 38,000 attendees and more than 16,000 retail companies from 99 countries, but also welcomed Microsoft CEO Satya Nadella, Best Buy CEO Corie Barry, and Starbucks CEO Kevin Johnson. Heavy guests.
NRF2020 site, the world’s largest retail exhibition
As one of the focus of retail technology in recent years, smart retail technology has attracted much attention. During this exhibition, as a new AI retail solution provider, Marlon Technology not only displayed on-site AI retail solutions including RetailAI asset protection, but also appeared on the booths of well-known companies such as Dell, Microsoft, and NVIDIA. Or on the cooperation list.
Nuggets overseas unmanned retail market, Marlon Technology showcases three major retail solutions
Code Long Technology’s solutions for the retail industry
At this conference, Marlon Technology showcased three major retail solutions, including the RetailAI asset protection solution, which focuses on helping retailers to significantly reduce cargo damage.; RetailAI intelligent container solutions that can effectively improve the user experience; and RetailAI intelligent weighing solutions that significantly increase weighing efficiency and effectively control cargo damage.
In summary, compared to the unmanned retail solutions in the market in recent years, these three solutions directly face the pain points of retailers and have a clear ROI (Return on Investment). Customers can use it at scale.
For retail customers with low profit margins, how to obtain the largest ROI has always been their main focus, especially in developed countries where labor costs are relatively high. Self-service checkout counters that have already begun to be installed are an important measure. one. According to a third-party survey by P & S Intelligence, due to increasing labor costs and in order to continuously optimize the consumer shopping experience, the global self-checkout counter market will reach $ 710 million in 2024, with a compound annual growth rate of 10.9%.
In recent years, as machine vision-based unmanned retail solutions have gradually been optimized, going overseas has also become an important choice. Taking unmanned retailing as an example, overseas is already a relatively mature concept. According to a third-party survey by P & S Intelligence, due to increasing labor costs and in order to continuously optimize the consumer shopping experience, the global market size of self-checkout desks will reach $ 710 million in 2024, with a compound annual growth rate of 10.9%.
But at the same time, the self-checkout desk also brought a new pain point-cargo damage. The data shows that in 2018 alone, the US retailer’s cargo damage reached US $ 4.7 billion. Among the self-service settlement links, some customers will have unscrupulous behaviors such as missing goods and replacing product barcodes, which will cause cargo damage to retail companies.
Code Long Technology debuts NRF 2020 with RetailAI solution
Solving the problems in this scenario has also become the most important entry point for Sialon’s solutions. After deploying the RetailAI asset protection solution, under the condition of protecting the personal information of customers, the self-checkout counter can accurately detect errors and missed scans in the settlement process and send reminders to store staff, effectively recovering the corresponding losses.
Salon Technology told 36. This solution has already landed in dozens of Fortune 500 retail stores. Actual data from the store operations show that in the supermarkets with more than one million SKUs, the RetailAI asset protection solution can reduce the settlement and scanning of goods by more than 80%.
Based on the same underlying AI product identification technology, Marlon Technology showcased the RetailAI intelligent weighing solution. By adding an ordinary camera, the RetailAI intelligent weighing solution can transform ordinary self-service vegetable and fruit scales into intelligent weighing platforms, giving it an automatic identification function, and supporting the identification of more than 300 categories of vegetables, fruits, and impurities. Covering more than 1 million SKUs, it can maintain a high level of accurate identification even when the product is bagged, which can help retailers better save labor costs and ensure asset safety, while providing customers with smoother and more efficient Friendly friendly weighing experience.
Salon Technology said that based on hundreds of thousands of usage data of Shenzhen’s local commercial supermarkets, the solution can achieve an identification accuracy of about 99%; compared with traditional self-service vegetable and fruit scales, customers Weighing time has been reduced by more than 60%-Previously, the average weighing process took 15 seconds, and the average use time was reduced to 5.5 seconds after using the smart solution. Previously, more than 65% of customers needed self-service weighing to help shop assistants. In the stores that use the RetailAI intelligent weighing solution, store assistant assistance has been reduced to less than 10%, and there is a continuous downward trend. At the same time, the RetailAI intelligent weighing solution can also provide retailers with efficient consumer behavior analysis tools and effective monitoring. Cargo damage, reducing the corresponding cargo damage rate by 50%.
In addition, Marlon Technology showcased the RetailAI smart container solution. This solution can use AI product identification technology to complete the intelligent transformation of traditional containers, realize the “snap and go, automatic settlement” smart shopping process, and provide consumers with a better experience at a lower cost. At present, Marlon Technology has reached in-depth cooperation with Haier for mass production, and has signed a number of brands and operators such as Mengniu, Junlebao, Coca-Cola, and plans to continue to expand its position and market in 2020.
Compared with smart retail ROI, Sialon Technology has backed up five years of technology accumulation
On February 12, 2019, Deloitte released its “Global Retail Power 2019” report. The report shows that in fiscal 2017 (the fiscal year ending June 2018), the revenue of the world’s top 250 retailers totaled $ 4.53 trillion, a compound growth rate of 5.7%, but the compound net profit margin was 3.2% from the previous fiscal year. decrease to2.3%.
E-commerce competes to diversify sales volume, high labor costs drive down profits, theft is difficult to prevent and increase losses, and the retail industry is facing internal and external problems. The artificial intelligence technology represented by machine vision is also considered as an important tool that is expected to change the offline retail situation.
Starting from product identification and value such opportunities, in 2017, Codeon Technology also gradually shifted its business focus from multi-industry business in parallel to focus on product identification applications, especially the retail industry. In the past three years, through continuous iteration and optimization, the current three major solutions have been formed. In fact, self-checkout machines, unmanned containers, and unmanned retail stores are also the three main directions of the current AI retail solution providers. The entry point of Marlon Technology is not only in line with the mainstream trend, but also has a unique entry point that focuses on landing. .
For retail customers, low-cost and rich product SKUs to ensure the accuracy of product identification are the basis for the commercialization of AI retail solutions. Judging from the results, Marlon Technology ’s RetailAI smart container solution currently supports merchants to self-service new, as long as 22 photos are taken in the app as required to complete the new process. In addition, in terms of the accuracy of product recognition, static recognition is the main, combined with dynamic recognition, and a 99% accuracy commercial recognition engine is built on the premise of ensuring low cost. This has also become one of the most important competitiveness of the company’s products.
The reason why this can be achieved is related to the technical investment of Malon Technology in the past few years. For example, many companies in the industry currently perform product recognition mainly through algorithms such as deep learning, which requires a large amount of accurate labeling data. In fact, in the retail industry, many customers often have a large amount of industry image data, but most of these data are labeled inconsistently, and some are not labeled or labeled incorrectly. For such noisy data, traditional deep learning algorithms need to spend a lot of money and time to clean the data, and the labor cost also increases significantly.
To this end, the technical team of Marlon Technology explored a training strategy for noise-containing data, introduced a weak supervised learning algorithm and a course learning algorithm, and designed a new algorithm strategy that can effectively suppress the negative impact of mislabeling , Has better generalization capabilities. In July 2017, at the time, the first CVPR2017 large-scale computer vision WebVision challenge co-organized by Google Research Institute and the Federal Institute of Technology in Zurich was based on the latest “weakly supervised learning” research results. 94.78%, which is comparable to the manual recognition rate of 94.0%-94.9% for performing the same task, and it is ahead of the second place with a clear advantage.
Large overseas market, Marlon Technology joins hands with well-known partners
As the oldest and most traditional line in historyIndustry, the retail industry has formed a relatively complete industrial chain and pattern. As an emerging AI technology vendor, to have better business results, it is often necessary to get on the “Express Train” of well-known partners.
Working with well-known partners with complementary advantages and launching high-ROI commercial products together is also one of the key business strategies of Marlon Technology. For example, in the direction of smart retail cabinets, Marlon Technology has reached cooperation with cabinet machine brands such as Haier and Aucma to launch smart refrigerator solutions together. There are preliminary mass production plans; in overseas markets, Marron Technology and the world’s largest self-service The settlement desk has reached cooperation with the corresponding hardware manufacturers to create and promote solutions for merchants to prevent cargo damage.
Today, giant companies have mastered channel resources and built mature software service platforms and hardware service systems. In more vertical segments such as retail, the choice of giant companies has also become a booster for the growth of emerging solution providers. For giants with customers, channels, and resources, they will also pay more attention to and select partners with more industry-specific insights and the ability to solve specific problems.
In the past few years, Marlon Technology has reached partnerships with well-known companies such as Accenture, Microsoft, NVIDIA, and Dell. In August 2018, Marlon Technology obtained Accenture’s strategic investment and reached an AI strategic cooperation with Accenture, becoming the first company Accenture invested in China. With Accenture, Marlon Technology will also have the opportunity to truly serve global retailers as part of the solution. In cooperation with Dell, based on the NVIDIA EGX platform to build real-time end-to-end machine learning pipelines in retail stores and supermarkets, to achieve edge computing deployment, will also help applications in global retailers. The success of Microsoft ’s “Gold Partner” identity certification and the successful selection of the NVIDIA Inception Premier upgrade acceleration plan will also help companies gain recognition in the market, especially overseas markets.
At the NRF 2020 retail show, Marlon Technology and its partners also joined hands to participate-participating in the Dell booth display, and brand exposure in the Microsoft booth. In addition, during the exhibition, Marlon Technology and Nvidia jointly issued a cooperation statement to jointly implement artificial intelligence product identification solutions to a wider market.
Every technological change not only means new opportunities for traditional players, but also often spawns a number of start-up companies with technical strength. With the entry of technology giants, retail manufacturers, and start-ups, the penetration rate of smart retail has gradually increased. In June 2019, an international research and consulting company IDC surveyed 400 retailers worldwide. The results showed that 28% of retailers have started to test or adopt unmanned settlement technology. As a result, AI asset protection solutions have great potential Landing scene. In the future, there will be more retail scenarios that require technological empowerment and transformation.
On February 12, 2019, Deloitte released its “Global Retail Power 2019” report. The report shows that in fiscal 2017 (the fiscal year ending June 2018), the revenue of the world’s top 250 retailers totaled $ 4.53 trillion, a compound growth rate of 5.7%, but the compound net profit margin was 3.2% from the previous fiscal year. decrease to2.3%.
E-commerce competes to diversify sales volume, high labor costs drive down profits, theft is difficult to prevent and increase losses, and the retail industry is facing internal and external problems. The artificial intelligence technology represented by machine vision is also considered as an important tool that is expected to change the offline retail situation.
Starting from product identification and value such opportunities, in 2017, Codeon Technology also gradually shifted its business focus from multi-industry business in parallel to focus on product identification applications, especially the retail industry. In the past three years, through continuous iteration and optimization, the current three major solutions have been formed. In fact, self-checkout machines, unmanned containers, and unmanned retail stores are also the three main directions of the current AI retail solution providers. The entry point of Marlon Technology is not only in line with the mainstream trend, but also has a unique entry point that focuses on landing. .
For retail customers, low-cost and rich product SKUs to ensure the accuracy of product identification are the basis for the commercialization of AI retail solutions. Judging from the results, Marlon Technology ’s RetailAI smart container solution currently supports merchants to self-service new, as long as 22 photos are taken in the app as required to complete the new process. In addition, in terms of the accuracy of product recognition, static recognition is the main, combined with dynamic recognition, and a 99% accuracy commercial recognition engine is built on the premise of ensuring low cost. This has also become one of the most important competitiveness of the company’s products.
The reason why this can be achieved is related to the technical investment of Malon Technology in the past few years. For example, many companies in the industry currently perform product recognition mainly through algorithms such as deep learning, which requires a large amount of accurate labeling data. In fact, in the retail industry, many customers often have a large amount of industry image data, but most of these data are labeled inconsistently, and some are not labeled or labeled incorrectly. For such noisy data, traditional deep learning algorithms need to spend a lot of money and time to clean the data, and the labor cost also increases significantly.
To this end, the technical team of Marlon Technology explored a training strategy for noise-containing data, introduced a weak supervised learning algorithm and a course learning algorithm, and designed a new algorithm strategy that can effectively suppress the negative impact of mislabeling , Has better generalization capabilities. In July 2017, at the time, the first CVPR2017 large-scale computer vision WebVision challenge co-organized by Google Research Institute and the Federal Institute of Technology in Zurich was based on the latest “weakly supervised learning” research results. 94.78%, which is comparable to the manual recognition rate of 94.0%-94.9% for performing the same task, and it is ahead of the second place with a clear advantage.