In industrial production areas, wrinkle detection is an important quality control method in sectors such as textiles, packaging, automotive, etc. Wrinkles can spoil the aesthetic appearance of products or affect the functionality and quality of the products. This situation can significantly reduce efficiency in production.

As the GoLive-R&D team, we continue to develop solutions by closely monitoring the problems in the industry. One of the solutions we have developed in this context is the Wrinkle Detection Application, created to prevent the negative situations caused by wrinkles in products.

What is Wrinkle Detection Application?

The Wrinkle Detection Application has been developed to detect and identify wrinkles or folds on an object or surface, determining their presence and location for reporting purposes. Wrinkle detection is carried out using image processing and artificial intelligence technologies.

The Wrinkle Detection Application was first implemented in the Automotive sector and has been used to detect wrinkles that may form on car seats.

Why is this application needed?

Wrinkles are unwanted and need to be avoided in the production process of seats. Prior to the Wrinkle Detection Application, the detection of seat wrinkles was a process left to human discretion. This led to the standards of the process being unclear and subjective. Supported by artificial intelligence, this new application has eliminated this ambiguity by establishing a standard for wrinkle errors. Moreover, by ironing only the areas with faults during the ironing process, time is saved during the seat ironing process, thereby increasing production speed and efficiency.

What is the Hardware Used?

Within the scope of the project, two color (RGB) cameras, one depth camera, and one Panel PC were used. Images of the back and seating areas of the seat are obtained through the color cameras, while the overall view of the chair is captured with the depth camera. The images received from the cameras are processed on the Panel PC, and a wrinkle control process is carried out and reported using an artificial intelligence model.

How is it working?

The wrinkle detection process includes these steps:

• Data Collection and Processing : After field installations are made, Seat images are collected for training the Artificial Intelligence model.
• Production Line Integration : The communication infrastructure on the production line with the Panel PC is established to trigger cameras and capture images when the seats on the production line enter the quality control area.
• Wrinkle Detection
• Reporting

Data Collection and Processing

After the field setup was completed, images of car seats were collected, and a training set was created from these images. The seat surface was divided into different areas, and segmentation models were used for this area division. Wrinkled areas in the dataset were specially labeled, and the model was trained with this dataset. Object detection models were used for wrinkle detection.

Production Line Integration

The seats on the production line trigger cameras to capture images when they enter the quality control area, creating a communication infrastructure with the Panel PC on the production line.

Wrinkle Detection

The images obtained from the cameras are fed into a trained object detection model. The trained model detects wrinkles on the car seat and the areas with wrinkles are marked with a red frame and reflected on the visual interface. After the wrinkles are detected, the wrinkle information of the relevant areas is recorded in the database.

Reporting

Using the information recorded in the database, tables are created from data such as heatmap images based on the density of wrinkles and average number of wrinkles in regions. These tables and images are used to visually summarize statistical results and are recorded in PDF format.

Flexibility and Control

Users can prevent mistakenly marking small wrinkles as defects by setting a certain wrinkle level threshold in the interface.

What Did The Application Bring Or Provide?

In the company where the application was installed, a pioneering step has been taken in the automotive seat manufacturing industry by integrating artificial intelligence and image processing technologies to increase quality and efficiency. This innovative approach has not only improved the production process of the company but also raised the quality standards in the automotive sector.