Quality inspection workers are being promoted to become data annotators and “teachers” of AI.
In the workshops of Hengshen Holding Group, the quality inspectors are no longer front-line textile workers, but AI quality inspectors. If the product fails to pass the inspection, there is no need for manual inspection. The industrial camera deployed at the front end can be used to scan and tell.
“In the past, the company’s quality inspection workers mainly adopted the ‘eyes + flashlight’ working mode, which was not only inefficient, but also the eyes may be damaged after working for a long time. Now, the company has cooperated with Baidu to build a visual After the detection system was ‘installed’, the ‘AA rate’ of the company’s products increased by nearly two percentage points.” At the “Textile Industry Artificial Intelligence High-Quality Development Forum – Quality Inspection Technology Special Topic” held in Shengze Town, Suzhou City, Jiangsu Province, Xiao Weimin, director of the information department of the chemical fiber sector of Hengshen Holding Group, said happily that the quality inspection problems that have plagued the company for many years have been basically solved.
Quality inspection problems need to be solved urgently
Nowadays, “big factories” have gradually become effective “foreign aid” for textile enterprises.
In addition to Baidu, Alibaba has developed AI algorithms in recent years that can complete quality inspection of mainstream fabrics such as chemical fiber, cotton, denim, and leather. Huawei has also created a joint smart textile inspection solution to serve the upgrade of the full-process automated production line for cloth quality inspection…
Obviously, the goal of “big factories” to “go away from the virtual and move towards the real” is focused on quality. Check on.
This choice is not surprising. Because the quality inspection link is one of the most urgent problems for textile companies.
“Every time during the peak production season, major companies will expand production capacity and seize market opportunities, but when it comes to the final quality inspection process, they often get stuck.” Xiao Weimin said that the textile industry has relatively high requirements for product quality. high. For example, the inspection content of fabric products usually includes the color, measurement, weight, warp and weft of fabric products, fabric structure, etc. “In the past, the common random inspection amount was 5%-10% of the total amount, but if the value of the fabric is high, or if the customer needs it, the factory will also conduct 100% inspection of the fabric after passing the random inspection.”
This poses considerable challenges to the professional abilities and work proficiency of quality inspection workers. But the difficulty doesn’t end there.
Generally speaking, production line workers in the textile industry are relatively concentrated in the quality inspection process, accounting for about 10% of the total number of workers. However, the working environment of textile enterprises is often faced with noise, dust and other interference, and many young people avoid it. Therefore, the quality inspection process is often carried out by workers in their forties and fifties.
“The quality inspection link of the factory has entered a difficult situation of being out of business and having no successors.” Xiao Weimin pointed out that if there is no stable supply of workers, the factory will be unable to provide reliable and qualified products to the market and cannot satisfy the market. demand, which in turn affects the production efficiency of the factory.
When manufacturing companies are in urgent need of digital transformation, the integration of digital technology and the real economy has begun to accelerate. The exploration of artificial intelligence by “big factories” is no longer limited to technological breakthroughs in computing power, algorithms, data, etc., but has gradually deepened from the perspective of industry applications and needs, starting from the quality inspection link, and exploring cooperation with textile enterprises. A “two-way” revolution.
Reinforcements from “big factories” are gradually increasing
However, as a supplier of technical solutions in the textile field, “big factories” were once quite embarrassed to have contact with enterprises. Many people believe that they do not understand manufacturing and scenarios, and the technology cannot be implemented. Although the cases are perfect, their practicality is low.
In this regard, Huang Feng, deputy general manager of Baidu Intelligent Cloud Smart Industry Division, said that Baidu has never meant to be independent from the manufacturing industry or work behind closed doors from the beginning, but has emphasized the need to go deep into reality and Let engineers “sink” to the front line to understand the actual situation and understand the company’s pain points. “Textile companies’ demands for improving efficiency, reducing costs and increasing quality have put forward higher requirements for information technology and solutions, which has also led Baidu to further think about how to use new generation information technologies such as artificial intelligence to help companies solve practical problems.”
“Baidu Intelligent Cloud engineers must stay on the production line, observe the entire quality inspection process, and communicate details with staff to achieve replicable experience, and perform model tuning based on data in real time to reduce the false detection rate. .” Nie Lei, head of Baidu Intelligent Cloud Industrial Vision Algorithm, also said that when working on some key projects of textile companies, the corresponding technology development team would stay in the factory for half a year or even longer. There is only one reason. Contact with applications in business scenarios will make you more down-to-earth.
The efforts of “big factories” have been recognized by textile companies. Xiao Weimin pointed out that the application of the appearance visual inspection system built by Baidu effectively solved the pain point of manual inspection being unable to maintain inspection quality consistency, and achieved centralized product appearance inspection through machine deep learning, AI technology, image recognition technology and 5G network transmission. Integration, thereby improving the product qualification rate.
“After the workshop introduced Baidu visual inspection technology, the labor efficiency alone has increased by 70%.” Li Feng, marketing director of Anqing Fengzhao Printing and Dyeing Co., Ltd. revealed that after the system was put into use, the workload of workers was greatly reduced. Pressure, the product quality rate is expected to increase from 95% to more than 98%, and the industrial value and economic benefits will be significantly improved.
In addition, according to a reporter from China Textile News, Alibaba’s AI algorithm has also been launched in many textile and garment factories across the country, helping factories to automatically complete quality inspection of raw materials, gray fabrics, finished fabrics, and garments in the entire production process.Work. Relevant companies generally report that the recognition accuracy is over 90%, far exceeding the manual level, and the overall efficiency has been greatly improved.
AI quality inspection is just the beginning
The “IEEE Global Survey: The Impact of Technology in 2022 and Beyond” released by the Institute of Electrical and Electronics Engineers (IEEE) shows that more than 80% are affected by Respondents believe that in the next five years, 1/4 of the world’s jobs will be replaced by artificial intelligence. Chinese respondents have higher expectations for “human-machine collaboration”: 70% of respondents believe that artificial intelligence will assist humans in completing more than half of the work in the next 10 years.
For the textile industry, AI quality inspection is obviously just the beginning of industrial intelligence. Yang Na, Director of Baidu Intelligent Cloud Textile Industry Solutions, said that in the future, Baidu will further deeply integrate AI technology with industrial models to help textile companies provide technical solutions such as energy consumption management and control optimization and resource allocation optimization, and help the textile industry create a transparent, intelligent and A predictive supply chain strives to drive the textile industry from a closed system to an open value ecology and achieve a comprehensive upgrade of the industrial chain.
In August this year, six departments including the Ministry of Science and Technology jointly issued the “Guiding Opinions on Accelerating Scenario Innovation and Promoting High-Quality Economic Development with High-level Application of Artificial Intelligence”, which specifically stated: “It is necessary to promote artificial intelligence and physical entities Deep economic integration is the main line.” This undoubtedly laid a stronger foundation for the cooperation between textile enterprises and “big factories”.
Qiao Yanjin, director of the Productivity Promotion Department of the China National Textile and Apparel Federation and director of the China Textile Information Center, pointed out that the current digital transformation of the textile industry is moving from single links and subdivided areas to an all-round industrial ecology. In the “14th Five-Year Plan”, the China Textile and Apparel Industry Federation emphasized that it is necessary to continuously strengthen the in-depth integration of key technologies in intelligent manufacturing applications such as the industrial Internet, big data, artificial intelligence, industrial robots, and blockchain in the textile industry, and to enhance the digitalization and Intelligent basic capabilities, the use of advanced applicable technologies and scientific management tools, accelerate the digital transformation and improvement of enterprises.
Qiao Yanjin said that building an open industrial ecosystem in the future will be a strategic path to accelerate the deep integration of artificial intelligence and the textile industry. Textile enterprises should take the initiative to build a new development pattern and strive to promote high-quality development.
</p
