Innovative start-up company with limited liability, “DigitaLLeather doo” was supported during the 10th cycle of the Early Development Program by the Fund for Innovation Activities of the Republic of Serbia (https://www.inovacionifond.rs/cir/program/program-ranog-razvoja). The name of the approved project was “Workstation for detection of defects on leather cut-parts – 2413”. The goal of this project was the creation of a workstation based on artificial intelligence, for the purposes of quality control of industrial leather. In accordance with the innovation project, the innovation start-up DigitaLLeather doo was founded on November 16, 2021.
The process of controlling the quality of industrial leather is carried out manually throughout the world, that is, the entire leather industry on a global level. The industrial leather quality control operator visually and tactilely inspects the entire industrial leather or leather sections. Based on knowledge or experience, he makes a conclusion about the presence of defects on the whole skin or skin section. Then he checks the defect tolerances using measuring devices, so depending on the checked quality, he rejects or lets the leather product continue in the process. This process has been traditionally applied in the leather industry probably since the founding of the leather industry. The problem with this approach is the reliability of the process, which is completely dependent on the quality control operator, i.e. the human factor. For this reason, there is a reduction in quality for customers and end users of industrial leather products. Also, a big problem is the reduction in the use of industrial leather, which directly affects the increase in costs and waste in the process. Omissions on the part of the industrial leather quality control operator lead to the need for re-production of certain parts and finishing of finished products, which additionally increases the costs and time of manufacturing the product, and certainly has negative consequences on the quality of the product. From which it can be concluded that this phenomenon can and does lead to a decrease in competitiveness on the market.
Our idea was to create a prototype of a technical solution, based on the application of artificial intelligence, with which we would reduce the influence of the human factor in the quality control process of leather scraps. The idea originated in a company that is a supplier of leather products for the automotive industry, which gave us additional insights into the process itself and the quality control procedure of leather scraps in industrial conditions. Through painstaking research of the market and competition, we came to the conclusions why there is still no stable solution on the market and decided to be pioneers in the application of artificial intelligence for detecting defects on industrial leather in Serbia and the region. Our idea was to digitize the knowledge and experience of industrial leather quality control operators, who work in a company that already has domain knowledge and experience, and has ordered stable quantities of industrial leather products that are serially produced.
The project “Workstation for detection of defects on leather cut-parts – 2413” lasted a year and resulted in the creation of a technical solution based on artificial intelligence, which we named “Prototype P1-22”. When testing the solution in industrial conditions, satisfactory results were achieved in the detection and classification of defects, within the appropriate time frames of individual control of cut leather parts. The technical maturity level of the developed prototype is TRL 5. The P1-22 prototype consists of hardware and software. The hardware is made in the form of a light chamber, square-shaped, semi-open type, in which parts of cut industrial leather are placed and with the help of a camera and appropriate lighting, the leather section is digitized. The software part of the solution consists of an API based on artificial intelligence, which is used to detect and classify defects if they are detected on a skin clip. The entire technical solution is also supported by a software web application for product data management, with which users and other interested parties can be informed in a timely manner about the performance achieved by the specified industrial leather quality control process.
