On December 12, 2023, a salon brought together elites from academia and industry was successfully held at the Radisson RED Shanghai Wujiaochang Hotel. The salon, hosted by Kaiyu Group and Tongji Consulting, specially invited experts and scholars from Tongji University to participate in the cutting-edge theme of "Core Edge, Intelligent Business Visual AI and Application of Large Models in Shopping Mall Operations". Intel and other partners also provided strong support for the forum.
Ying Xiangxiang, general manager of Kaiyu Digital Technology, made an opening speech, Deng Songlin, head of Fenglan International Project of Kaiyu Group, shared the application of visual AI and ** large model in the operation of Fenglan International Mall, Shao Wenjian, chief engineer of Intel's Network and Edge Computing Smart City Division, gave a speech entitled "Visual Large Model Construction Edge Training and Pushing Integrated Solution", and Ma Shuangbao, researcher of Shanghai Collaborative Innovation Center for Artificial Intelligence Social Governance, shared the Observation on Face Recognition Empowering Public Safety and Public Service Acceptance: A Social Experiment Based on a Street in Shanghai.
In the roundtable discussion session, Ying Xiangxiang, general manager of Kaiyu Digital Technology, as the moderator, had a wonderful dialogue with the guests: Zhao Yudi, CTO of Kaiyu Group, Xiao Chao, teacher of the School of Economics and Management of Tongji University, Shao Wenjian, chief engineer of Intel's Network and Edge Computing Smart City Division, and Deng Songlin, head of Fenglan International Project of Kaiyu Group, had a wonderful dialogue on visual AI and large model applications.
A number of industry experts gathered at the salon, aiming to strengthen industry-university-research exchanges and promote the continuous improvement of the commercial real estate industry. In-depth discussions were held on topics such as customer practice, technical insights, and academic frontiers of visual AI and large models in the operation of shopping malls, and how to make more effective use of visual AI and large models to create a digital and intelligent operation model for shopping malls.
The traditional business format has been impacted, and the transformation is imminent
According to the statistics of the China Chain Store & Franchise Association, by the end of 2022, the stock of domestic shopping center projects was close to 6,700 (more than 30,000 square meters), and the overall development and expansion of the shopping center market has slowed down. In addition, the homogenization of shopping mall formats, brands, and marketing methods is serious, and the industry is highly involutedThe upgrade of traditional intelligent analysis is faced with the high cost and long cycle caused by the need to replace existing camerasThe refinement of data operation is insufficient, and consumer behaviors and habits need to be deeply explored. In the era of stock, how to break out of the encirclement has become a difficult problem.
Kaiyu Group prescribes the right medicine to empower the traditional economy
In the wave of digital transformation, Fenglan International Shopping Center chose to cooperate with Kaiyu Group and Intel, which are advanced digital technology solution providers in China.
Through the cooperation with Kaiyu Group and Intel, Fenglan International Shopping Center quickly realized the digital upgrade of the existing security monitoring system. Since the Intel**AI Compute Box has good compatibility, it can be used inIn a short period of time, the existing security surveillance cameras can be reused, avoiding waste and tedious work of replacing equipment. This efficient modernization process is not only thatTime and cost savingsIt is also an application that pushes traditional security monitoring to a higher level.
Based on Intel**AI Compute Box (Video AI Box), the digital solution of Kaiyu Fenglan International Shopping Center includes the following core functions and value applications:
More accurate and rich insights into passenger flow. On the basis of traditional passenger flow, with the help of open-domain cross-mirror humanoid trajectory tracking algorithm and REID technology, not only the statistics of the number of passengers, but also the statistics of the whole passenger flow in the dimensions of deduplication and batch can be realized. In addition, through insight into each key node such as the passenger flow at each entrance, floor, and store, the analysis dimensions include customer portrait, travel time, heat and circulation line, business format association, and climbing rate. The data is abundant, and the level and depth are taken into account.
Through in-depth insight into passenger flow data, it analyzes the preferences of passenger flow for different business formats, and provides refined operational data for various decisions such as intelligent renting, business combination, store layout, and marketing strategy. And based on AI algorithms, it analyzes customer portrait data to provide a basis for deepening differentiated marketing methods. In addition, based on the Owl Visitor Statistics app, the software will help malls better empower brands, enabling brands to grasp the effectiveness of pop-up store activities more efficiently and accurately, so as to make more informed decisions.
Intelligent AI inspection. Through the use of existing security monitoring, we also apply the ability of AI inspection to the project. At present, the intrusion identification of key entrances and exits and jewelry stores has been deployedOnce there is an intrusion in these areas during non-business hours, the system will automatically trigger an alarm event and notify the relevant staff through SMS, APP, and email. In this way, we can respond to and deal with any anomalies in a timely manner, thereby increasing the level of security throughout the mall and ensuring the safety of property.
In the future, we will continue to work with customers to explore more applications of AI inspection in shopping malls." For example, we will introduce AI to monitor falls, fights, and inappropriate behaviors such as smoking. Through the empowerment of visual AI, shopping malls will become more intelligent and provide customers with a safer and more comfortable shopping environment.
AI retrieval based on large models. In shopping malls, we often encounter the problem of getting lost due to parental negligence or naughty children. In the past, we could only rely on manual viewing of the surveillance** to find the lost child, which required us to first determine the approximate time and place of the child's loss, and then find the corresponding surveillance**, and rely on the manual to carefully review frame by frame, which is very time-consuming and laborious.
Then do we have a more efficient way, the emergence of the first large model, in the future can help Fenglan International solve this problem, only need to enter the characteristics of the corresponding target, such as gender, clothing color, etc., the large model can quickly feature the trajectory of the person and the relevant ** screen to call out, which can simplify the search workload of Fenglan International, improve the ability to deal with emergencies.
In the future, we hope to help shopping malls find lost children and lost items more efficiently through the application of large models, as well as gain insight into whether security guards follow the established route and so on.
Advantages of the digital solution of Kaiyu Fenglan International Shopping Center based on Intel's **AI Compute Box (Video AI Box):
Efficient recycling。Based on the design of scheme compatibility, 90% of the existing security monitoring of Fenglan International Shopping Center can be completed within one week of the implementation, deployment and commissioning of the project, which greatly improves the efficiency of digital upgrading and transformation, reduces the cost of transformation, and pushes the traditional security monitoring to a higher level of application.
Passenger flow insights are abundant。It not only provides insight into the number of passengers, but also outputs data in more dimensions such as the number of people, batches, entrances and exits, floors, formats, stores, trajectories and heat, customer portraits, stay duration, relevance, and climbing rate, helping Fenglan International improve its refined operations.
Powerful automated processing capabilities。**The large model has powerful automatic processing capabilities, which can greatly improve the efficiency of algorithm customization, reduce the workload and time cost, and thus improve the ability of the shopping center to deal with emergencies.
Strengthen digital consumption and help industrial upgrading
With the iteration of consumer consumption habits and the new era of digital consumption, digital operation and intelligent consumption experience will become a new battlefield for the stock competition of shopping malls. With its own accumulation and precipitation in digital transformation, Kaiyu Group provides another example for the digital transformation of shopping malls and the construction of smart business districts.
Shopping malls play an irreplaceable role in stimulating consumption, improving the quality of life, and solving employment. The digital solutions provided by Kaiyu Group for the traditional shopping center industry are based on Intel's **AI computing box and adopt the "cloud edge" system architecture. Moreover, backed by the latest generation of Intel Core processors and Intel A380 A770 graphics card-empowered AI computing box, through the deep combination of CPU and GPU, the independent graphics and core graphics work together to achieve a double improvement in the performance and efficiency of edge inference of open-domain algorithms. The high compatibility can be directly connected to the camera equipment of the shopping mall, which greatly saves the cost of digital transformation of the shopping center and improves the deployment efficiency.
At present, as a provider of digital solutions for the retail industry, Kaiyu Group has provided digital upgrade services for more than 3,500 partners. In the future, Kaiyu Group will work with partners to implement more technologies and empower commercial real estate to win the future in the stock era.