What is effective AI cooperation?Weishin is willing to appear in practice experience

The current situation is difficult, and the pursuit of cost reduction and efficiency will undoubtedly become an important proposition for enterprises.

At this node, if a technology is deployed, your development cost is reduced by 90%compared to the traditional way, the labor cost is greatly reduced, and the accuracy rate can be improved at the same time. Will you accept it?

If it is true and reliable, I believe that no one who has visionary business managers will refuse, and there is no reason to refuse.

Such opportunities are particularly valuable, but it also happens.

Veoskeng (Xiamen) Intelligent Technology Co., Ltd. (hereinafter referred to as Westken), an integrated integrator of the industrial quality inspection field, met EasyDL, a zero -threshold AI development platform of Baidu Flying Patriotic Enterprise. The latter's AI capability is applied to the car bearings production quality inspection scenario developed by the former. After verifying the running, the effect of reducing costs and efficiency is randomly displayed, exceeding expectations.

The recent influential consulting agency Kaijie issued a report called "Technovision 2022" ("Science and Technology Vision 2022"), which mentioned two views: "Our dependence on technology is exponentially growth." " The technical field continues to expand, enhance and adapt at amazing speed. "

Obviously, the industrial intelligent technology represented by AI has become the same as business, and considering the core element of a company's competitiveness and promoting the development of the enterprise, it has become the general trend.

The cooperation between Westken and the Baidu fly paddle EasyDL has become a typical typical of traditional enterprises in the process of embrace AI and digitalization. We try to make in -depth search and sharing. Essence

When Vexyin meets the Baidu flying paddle, it starts by accident but inevitable

How did Westa Ken contact the flying paddle?

"Coincidentally, it is recommended by our colleagues in Shenzhen when chatting."

Li Shuihan, the head of the Instrument Division of Weixing (Xiamen) Intelligent Technology Co., Ltd., is struggling to find the problem of detecting the problem of "grinding burns" of the car bearings: grinding burns is a bad phenomenon that workpieces occur after grinding processing. It belongs to a hidden defect that occurs on the surface of the workpiece. The difficulty of detection and the quality requirements of the workpiece are extremely strict. Even in the field of technology research and development, Li Shuihan still feels the arduous task.

"Colleagues have used flying paddle, let us try it too. We contacted the person in charge of the flying paddle in Xiamen to see if it can promote the landing."

Everything began with accidental recommendations, which contributed to the perfect start of the two cooperation, and eventually promoted the development of each other's expectations.

Behind this occasion, there is an inevitable ingredient, because of the policy, and the trend.

The Ministry of Industry and Information Technology jointly issued the 14th Five -Year Smart Manufacturing Development Plan for the National Development and Reform Commission, the Ministry of Education, and the Ministry of Science and Technology. The plan proposes that by 2025, manufacturing enterprises above 70%of the scale will basically realize digital networking, and build more than 500 intelligent manufacturing demonstration factories that lead the industry's development.

In addition, a report from Deloittes showed that 83%of the users said that they believed that artificial intelligence technology was one of the priority priorities in the entire digital process.

Baidu flying paddle, as the first self -developed, rich, open source industrial -grade deep learning platform in China, has served 180,000 enterprise institutions, which not only deeply meets the needs of enterprise application, but also has an active developer community ecology. It is through AI through AI The best choice to realize the intelligent transformation of the industry.

Westrin Kenkee has a strong custom AI model appeal. Although the own algorithm team has been established for nearly 1-2 years, it has invested nearly one million R & D costs, but the quality inspection direction of bearing accessories production is extremely high. In the end, it is limited by the effect of technology and algorithm development. Land on the production line.

In this way, a large amount of investment consumed the valuable time, and there was no corresponding output. The enterprise urgently needed timely to stop the loss and find effective solutions.

With high trial and error costs and accumulated experience, with their real needs, West Ken finally met and enabled the paddle to become an inevitable, and it can even be described as "meeting each other and hate".

What is effective AI cooperation? Weishinya shows out the saying

Bearing, it is everywhere in life: bicycles, cars, and electric toys cannot be separated from it. There are usually 30 types of bearing in a car, which are distributed in all positions of the car, which is related to the life and quality of the car.

The quality detection of bearing production is particularly critical. In the traditional way, each bearings test depend on people, hurting the eyes and energy cannot be concentrated for a long time. Essence

In the practice of AI, those who have not produced or have not been verified can be regarded as "paper tiger". May wish to see what kind of sparks have Westrockeng and Baidu fly paddle have been wiped out in the intelligent building of car bearings?

An old system integrator as an industrial quality inspection field serves giant companies such as Boeing, Airbus, and COMIC. The quality inspection business scenario is sufficient and requires new AI capabilities to bless innovation. Determined from an open source deep learning platform for industrial practice. The in -depth practical cooperation between the two decided to identify and resolve complex challenges through creative solutions.

In the visual direction, Westrin began to transform visual detection technology from the traditional vortex detection method. The technical capabilities based on the Baidu fly paddle EasyDL platform created the finished bearing visual detection system. The rich Baidu large -scale pre -training model and automation model optimization mechanism has been accumulated to achieve one -stop intelligent labeling, model training, and service deployment. Essence

It is easy to get started, the data labeling and model training are very smooth -soon, Westrin Ken creates a set of AI visual detection systems through Baidu fly paddle EasyDL, which can not only deal with complex and diverse bearing flaws, such as end surface and outer diameter scratch scratches More importantly, the problem of grinding burns that had been troubled before was solved well.

"The accuracy is higher than the previously tried methods." During the development and testing process, the most impressive impression of the Westken technical team was the AI ​​model accuracy implemented by Baidu EasyDL. "The models developed before often need to be tuned, but the defect detection system based on flying paddle can achieve a good effect on different types of defect detection after iteration."

In this way, whether it is to detect cracks, pores, looseness, hardness, soft dots, etc., or testing the size, form, appearance, and processing process of the components, this quality inspection is inspected for the test of the components. The system provides a strong guarantee, solve the raw materials of parts and components, and product quality problems that may exist during the processing, thermal treatment, and precision processing process, and provide safety guarantee for power systems and transmission systems.

In just three months, the Visual detection system of Westken's finished product bearing has achieved business launch. It can not only complete a test within 100ms, but also keep the error rate at about 3%, which truly achieves cost reduction and efficiency. "Compared with the traditional development method, our project uses EasyDL for model development, prompting the overall development costs to be reduced by 90%."

This set of bearing visual detection systems also successfully helped Westle to get the top ranking of the domestic bearing field and the seventh "Human Normal Group" in the world. Essence It can be seen that Westken has taken the pace of intelligent transformation and upgrading and has achieved further business expansion.

It's not done overnight, get the result in practice

Of course, even if the paddle has accumulated a lot of industrial intelligent landing cases and experiences, the process of cooperation with Weersan is not achieved overnight, and the verification exploration period from 0 to 1 has been experienced in the middle.

"We must know the whole process. Although we have done it for so long in the bearing industry, we have discovered a lot of new pits since cooperating with flying paddle. We have not stepped on it before. Shi Ken said frankly that it involves factors such as data quality and detection environment.

But why not? Compared with the previous one million invested water drifting and spending a long time, there was no way to meet the expectations. The cooperation with the flying paddle has really achieved achievements and achieved results.

"The paddle is really well done in this piece."

In the process of promoting, Westken also fully felt the sincerity and intention of paddle to empower the intelligent transformation and upgrading of industries. "The technical team has given great support, whether it is the communication and adjustment of various issues, or the customized development of demand."

In industrial practice, different scenes will bring different tests to the landing of AI. The two parties in depth of the industry scene digging problems and finding the essence are not only to solve their own problems, but also to accumulate more effective experience for the "exploration" model of the industry field.

Small and medium -sized enterprise transformation guidelines: hug AI and fly paddle in progress

At present, under the good results and achievements, Westken's finished bearing visual detection system developed based on the Baidu fly paddle EasyDL has landed in Xiamen, Ningbo, Qingdao and other landing.

The paddle is positioned as an industrial -level deep learning platform, and it is due to industrial practice. It is one of its important features. The cooperation with Westken also expresses this characteristic to the fullest.

Flying paddle is always committed to in -depth integration with the industry, working with many industry partners, empowering AI technology, anchoring the pain points and difficulties of the industry, and constantly forming high -quality available solutions. As of May this year, flying paddle has been widely used in industry, agriculture, service industry, etc., serving 4.77 million developers, creating 560,000 models, serving 180,000 enterprise institutions, and in industrial quality inspection, smart energy, smart cities, finance, finance, finance, finance In many fields such as medical care, many cases of industrial landing have been precipitated in many fields of fields.

Big Data and Analysis Resources TDWI pointed out that the organization is using AI/ML technology to make more wise decisions and improve competitive advantages. It said, "Artificial intelligence is the core of the digital revolution in the field of analysis." It "promises to help organizations improve operations and bring new income opportunities."

In the "2022 Science and Technology Vision" report released by Kaijie, Kaijie pointed out that data, artificial intelligence, and digital twin technology have been widely used in business processes. The process of new ideas has improved, and new models, new formats, and new roads.

The innovative development and evolution of enterprises depends on the pattern, depending on value, and think about it. From a larger framework and a wider perspective, it depends on what value to create for consumers.

- END -