Traditional textiles are in the "original stage". Is computer vision + intelligent inspection opportunity a good entry point?


With the continuous maturity of artificial intelligence […]

With the continuous maturity of artificial intelligence, especially in the industry based on computer vision technology is not a minority. At present, artificial intelligence has also been applied quickly and better in the fields of security and finance. In addition, more and more entrepreneurs are trying to bring new breakthroughs in related fields such as industry and manufacturing to reduce costs, improve efficiency, and even gradually replace traditional manual methods.

Taking the textile industry as an example, in the textile process, it can generally be divided into spinning, weaving, dyeing and finishing processes. Among them, the cloth inspection machine is an important part of the weaving process. The detection of fabric defects is also a key link in the quality control of fabrics by major textile mills. At present, the domestic textile industry still detects the defects on the original raw cloth produced in the manual inspection stage. The manual inspection machine often has many defects such as slow speed, limited production, high missed detection rate and poor continuity of detection. Therefore, it is urgent to develop a novel, fast and accurate method for automatic detection of fabric defects.

Among them, Dr. Bu from Shenzhen is a high-quality fabric trading platform using artificial intelligence inspection machines, currently mainly for fabric traders. In fact, founder Zhang Yi also said that Dr. Bu's predecessor mainly focused on the development and production of new type of cloth inspection machine based on computer vision and deep learning algorithms, but the subsequent discovery, for users who rely on manual inspection for a long time, now The level of acceptance is not high and it is difficult to generate willingness to pay.

In Zhang Yi's view, there are three main problems in the textile industry: one is that it is difficult to find a cloth, the channels are complicated, the information asymmetry leads to a large number of hackers, and the transaction cost remains high. Second, the inspection is difficult, and the cost of manual inspection is low. High, false detection rate, low efficiency and facing the shortage of labor; Third, the difficulty of buying cloth, the vicious competition of middlemen and manufacturers often bring about shortcomings such as lack of short code and shoddy.

To this end, on the original basis, Dr. Bu has made some adjustments in his business at this stage. On the one hand, quality inspection services are provided through intelligent inspection machines; on the other hand, fabric trading platforms are provided. It is understood that the difficulty in detecting defects in fabrics mainly lies in the variety of textures and morphological structures of the fabrics, especially the types of defects are very different, so it is difficult to identify all types of defects. Traditionally based on the contrast of fabric images, in view of the large amount of data processing, the recognition rate is low and the detection speed is also limited.

In order to solve this problem, Dr. Bu mainly through the accumulation and deep learning of defective feature points such as holes and scratches. From the actual effect, compared with the average manual inspection cloth speed of 0.2m / s, Dr. Bu intelligent inspection machine can improve the work efficiency by 5 times; in terms of recognition rate, within the textile industry, within 2 points Qualified products, 70% of qualified products can be inspected by traditional manual methods, and 87% of zero-accuracy can be achieved by Dr. Bu's intelligent cloth inspection machine; in terms of cost, one intelligent inspection machine can replace 8 traditional ones. The cloth inspection workers save about 500,000 yuan per year.

As for how to push to the market, Zhang Yi said that on the one hand, it will directly target customers through relevant exhibitions. On the other hand, the company will quickly establish an operation promotion team in its fabric base through authorized agents, and follow up each file in a carpet-style manner. Mouth and manufacturer and trading company. In terms of profitability, Dr. Bu will provide third-party inspection services for fabrics as the main source of income. The follow-up will mainly provide sales services for quality fabrics and traders' stock fabrics through the fabric trading platform to solve the difficulties of inspection and sale. The problem bonding fabric.

Regarding market competition, Zhang Yi also said that the typical participants in the market, such as the high-end fabric inspection factory, are still mainly implemented by manual inspection mode in the traditional order mode, with quality assurance but the efficiency cannot be improved quickly, and the technology The willingness to iterate is not enough to support it to bear a large development cost; while in the fabric e-commerce platform, the typical chain, such as the chain, currently adopts the “Taobao” model to serve fabrics, and the market size is high, but it is lacking. Quality assurance, for users to pay more and more attention to quality issues today does not apply; in contrast, Dr. Bu's core strength lies in the accumulation of algorithms and databases, that is, to solve existing manual problems, quality problems, and efficiency problems by artificial intelligence. .

On the team side, founder Zhang Yi is a serial entrepreneur and graduated from the University of New Hampshire computer science; co-founder Wu Junxi, who worked at Shanghai Minyue Textile Co., Ltd., has sold millions of dollars every year for five years and is now mainly responsible for Dr. Aspect work.

It is reported that Dr. Bu is currently in the stage of angel round financing, estimated to be 2 million RMB, which will be mainly used for team expansion and product research and development. It is expected that the smart fabric inspection factory experience store will be opened in China's five major fabric production bases in the coming year.