What is the shackle that restricts the large-scale landing of AI in cities? Is it hash rate, data, or cost?
These factors are indeed a problem facing AI companies, but for AI users, the most direct problem is how to quickly find the algorithm that suits their needs and achieve rapid deployment.
The AI industry also has a barrel effect. At present, AI has made great efforts in the fields of security, finance, transportation, and medical treatment, and has achieved remarkable results; but there are still a large number of small areas in the industry, and AI has not yet been able to achieve full coverage. It is these thousands of fragmented scenarios and problems that have become obstacles to the large-scale industrialization of AI.
For example, grassroots managers can use AI to achieve intelligent transformation of security systems, but managers also need to deal with problems such as occupied fire escapes, high-altitude projectiles, and overflowing garbage cans. These problems are trivial and complicated, but the AI market lacks similar long-tail algorithms, even if there are related algorithms, grass-roots managers lack technical background, it is difficult to achieve rapid deployment and rapid application.
For another example, after the implementation of the "Bright Kitchen Bright Stove" project, although it has been possible for catering service providers to show the situation of the back kitchen to the public in the form of video images, if the Market Supervision and Administration Bureau also hopes to achieve the reminder of irregular behavior in the back kitchen and the frequent occurrence of restaurant rectification of irregular behavior, it is necessary to deploy algorithms such as unwearing hat detection, workwear detection, smoking detection, and rat detection. In order to "put together" the required algorithms at one time, the market supervision and management department needs to find different manufacturers and channels, which is not only time-consuming and laborious, but also the application cost is high.
It is these trivial problems that have become the biggest obstacle to the landing of AI. It can be seen that if artificial intelligence is to penetrate deep into the industry and achieve larger-scale development, it is necessary to solve the long-tail application problem of AI.
In order to enable AI to penetrate into long-tail applications more efficiently and at low cost, Yuntian Lifei, the governing unit of Guangdong Artificial Intelligence Industry Association, has launched an AI algorithm service platform - "Deep Wisdom".
AI intelligent perception system based on "deep intelligence"
For digital cities, one-stop AI resources are matched
Deep Wisdom is a platform-based product for the digital city field, including algorithm warehouse, algorithm training, intelligent scheduling, data governance and other functional modules, which can provide users with key AI resources such as algorithms and computing power in one stop, realize the rapid deployment of AI in long-tail applications, and transform urban grassroots governance from manpower-intensive to technology-intensive, from empirical decision-making to data decision-making, and from passive response to active governance.
The platform has three core highlights:
1.N kinds of algorithms to meet user needs at one time
At present, Shenzhi can provide video image analysis, speech recognition, semantic analysis, text analysis 4 categories, more than 100 kinds of algorithms, including road damage, road waterlogging, riding electric vehicles without helmets, out-of-store operations, manhole cover abnormalities and other urban comprehensive management, community management and other fields of long-tail algorithms, can meet the cumbersome needs of grass-roots urban governance work.
A single flower is not spring, a hundred flowers bloom in the spring and fill the garden. In the face of the digital and intelligent transformation of the huge scene of digital city construction, the algorithm of Yuntian Lifei alone cannot meet the needs of a large number of long-tail scenarios. Therefore, Deep Wisdom is also open to third-party algorithm manufacturers, which not only provides a benign algorithm industry ecology for the majority of small and medium-sized algorithm manufacturers, but also meets the AI algorithm needs of more users for more long-tail scenarios.
If the algorithm provided by Deep Wisdom still cannot meet the needs of users, users can also train their own algorithms through Deep Wisdom. Users only need to access relevant data resources in the platform, the platform can help users complete the data annotation work, and provide algorithm training tools, so that users can obtain high-quality algorithms at low cost.
Deep Intelligence can provide hundreds of long-tail algorithms
2. Intelligent distribution and accurate recommendation, so that the business flows from flow
Whether it is online shopping or reading news, big data can always "guess" the user's preferences and make accurate recommendations according to our clicks and browsing time.
Shen Zhi's precise recommendation has the same magic.
First of all, with the support of a variety of algorithms, Shenzhi can intelligently analyze the problems existing in various scenarios in the city, and intelligently assign the problems to the corresponding departments according to the responsibilities of the departments, and remind the relevant departments to solve them.
For example, if it is detected that there is water on the road, the manhole cover is abnormal, etc., the problem is pushed to the urban management; if the motor vehicle is randomly parked, it is pushed to the traffic management department.
After assigning tasks to various departments, Deep Wisdom can also realize accurate recommendation of tasks according to the laws of past business processing.
For example, problems such as not wearing masks in public places, riding without wearing helmets, etc., are difficult to prevent and prevent in time, which is a non-urgent task for business personnel, so when business personnel receive similar situations, they will not deal with them immediately; but for dangerous situations such as abnormal manhole covers and fireworks alarms, business personnel will immediately contact relevant personnel to deal with them.
In the process of using Deep Wisdom to handle tasks, Deep Wisdom is also constantly "learning", which can learn the urgency of business and the scale of event disposal according to the efficiency of business department disposal feedback, and more accurately recommend effective events to help improve the efficiency of task processing of business departments.
3. Intelligent decision-making, so that empirical decision-making to data decision-making
In addition to being able to find problems and push them to relevant departments, Deep Wisdom can also assist relevant departments in decision-making.
Deep Wisdom can quickly summarize the laws of various problems and present the laws to the business department in the form of data to provide support for the decision-making of the business department.
For example, riding an electric vehicle without wearing a helmet has always been a headache for the traffic management department. People riding electric vehicles tend to "whistle by", and it is difficult to remind people who are not wearing helmets in time. But if left unchecked, there will be great security risks.
In the face of such situations, Deep Intelligence can find the law of problem occurrence on the basis of intelligent analysis and provide a decision-making basis for the business department.
For example, in the application of a certain place, Shenzhi found that the situation of riding without a helmet occurred on a certain road in the noon time period, and the traffic management department deployed personnel at the corresponding time and place according to this situation to remind the cyclists to wear helmets. Good results have been achieved in dealing with this issue.
Enhance the modernization of urban governance capabilities
At present, Shenzhen Wisdom has been launched in shenzhen longgang, shenzhen longhua, Foshan Shunde and other regional political and digital bureaus, as well as isolation hotels and other scenes, and has achieved good results in urban governance, epidemic prevention and control and other fields.
The first is to achieve the co-construction and co-governance and sharing of front-end resources.
In the past, in order to carry out digital and intelligent transformation work, various business departments often needed to build a complete system from terminal to cloud, which not only invested hugely in the early stage, but also easily formed data islands in the later stage.
However, after the application of deep wisdom, the political and digital bureau or the big data bureau can use the platform's algorithm to analyze the video image content on the basis of coordinating the city's video image data, and intelligently distribute the problems found to the business department. When business units need special algorithms, they can also be acquired or trained on deep intelligence, without the need to build or purchase equipment and systems separately.
In addition, through the unified scheduling algorithm, computing power, data and other resources, the data of each business department can be precipitated on the same platform, and the analysis of multi-dimensional big data can be realized in the future, so that all business departments can share the results of data analysis.
Second, it can greatly improve the operational processing efficiency of grass-roots staff and save a lot of labor costs.
If you rely on manual real-time viewing of video images, a person can watch about 20 videos at the same time, and in 8 hours of working time, about 10 events can be found from the video in a week and advance processing.
After realizing the unified scheduling and supply of AI resources through deep wisdom, the system can realize 24-hour uninterrupted monitoring of all-region video and push events in real time, liberating grass-roots staff from tedious and repetitive work.
Taking Shenzhen Longgang District as an example, Shenzhi can find tens of thousands of events every day, and on the basis of intelligent learning of the work needs of various business departments, an average of 1,000 events are pushed to the business departments every day, so that grass-roots staff can shift more energy from finding problems to solving problems.
The third is to drive the development of the digital economy and the AI ecosystem.
AI is the core driving force of the development of the digital economy, and Deep Wisdom can provide digital and intelligent transformation resources for a large number of long-tail scenarios in the city, and effectively promote the landing of AI in more fields and scenarios.
In addition, Deep Wisdom can provide zero-code, low-cost algorithm training platforms for the majority of small and medium-sized enterprises and institutions, so that more subjects lacking professional AI personnel can participate in the training and application of AI, and promote the development of the regional AI industry ecosystem.
In terms of application deployment, Shenzhi supports the intelligent application of cities and districts by building a central platform, and can also quickly meet the needs of intelligent application transformation in specific scenarios by deploying the lightweight method of all-in-one machines.
For example, in centralized isolation places such as isolated hotels, lightweight deep intelligence AI algorithm service all-in-one machines can be deployed to identify irregular epidemic prevention behaviors in isolation places and meet the needs of managers for business digitization and intelligence.
The core support of self-evolving urban intelligent body
The strategic goal of Yuntian Lifei is to create a "1+1+N" self-evolving urban intelligent body, that is, an intelligent and ubiquitous perception network, a self-learning and self-evolving urban super brain, and N AI-enabled intelligent applications.
The AI algorithm service platform "Deep Wisdom" is an important support for the urban super brain, which can meet the needs of various industries for AI technology through the scheduling of AI resources such as algorithms, computing power, and data, and allow AI to penetrate into various scenarios of the city.
Yuntian Life focuses on the two major areas of digital city and human settlements, and has landed a series of products and solutions.
Such as dynamic/static portrait system - "deep eye", multi-dimensional big data analysis system - "deep sea", video structured system - "deep extract", traffic OD analysis system - "deep", etc., as well as the business Qi series for commercial and book city scenes, and the Yuan Qi series for pan-park scenes.
AI algorithm service platform - "Deep Wisdom" is another "masterpiece" of Yuntian Life in the field of digital cities.
In the future, Yuntian Lifei will continue to move towards the goal of building a self-evolving urban intelligent body, give full play to the advantages of "algorithm chipping" and "end-cloud collaboration", make AI technology more extensive and in-depth serve the intelligent operation process of the city, and contribute to the vigorous development of the AI industry.
Introduction of Shenzhen Yuntian Lifei Technology Co., Ltd
As a leading artificial intelligence enterprise in the industry, Yuntian Lifei takes artificial intelligence algorithms and chip technology as the core, and has the ability to chip algorithms, creating a series of products and solutions for digital cities and human settlements. At present, Yuntian Lifei has implemented artificial intelligence technology, products and solutions in dozens of large and medium-sized cities and many projects such as Shenzhen, Shanghai, Chengdu, Qingdao and Hangzhou, empowering the intelligent upgrading of digital city operation management and human settlements.
At the level of AI algorithms, Yuntian Lifei comprehensively explores various fields of AI recognition algorithms, including dynamic/static portrait recognition, target attribute recognition, posture recognition, gesture recognition, vehicle recognition, etc., which are built by the international senior team that won the international visual competition championship and won the first place in a number of algorithm PK competitions.
At the level of AI chips, Yuntian Lifei AI chips have won national awards and undertaken national projects for many times. In addition to the neural network chip project of the Ministry of Industry and Information Technology, Yuntian Lifei also undertook the major project of artificial intelligence chips of the National Development and Reform Commission and the Ministry of Science and Technology. In 2018, Yuntian Lifei's second-generation independent intellectual property neural network processor chip DeepEye1000 was put into production, and in 2019, it was independently commercialized.