OpenMV
Shields
Lens
Robotics
Book
Video
Download
Docs
Forum
OpenMV.io
GithubIn ceramic production lines, defects such as bubbles, cracks, and glaze impurities on product surfaces are often difficult to detect with the naked eye, and manual quality inspection suffers from low efficiency and poor stability. To address this, ceramic manufacturers have adopted SingTown Technology's OpenMV smart cameras to build a machine vision-based surface inspection system, enabling real-time image recognition and defect localization on ceramic surfaces.
The system is suitable for the inspection of small ceramic products with solid-colour backgrounds and fixed positions, with an inspection distance of approximately 30 cm, enabling automated inspection and sorting on the production line.

Ceramic manufacturing enterprises have established a ceramic surface defect detection system based on the OpenMV smart camera. This system integrates multiple algorithmic modules, including computer vision, edge AI, image recognition, edge detection, color segmentation, and texture feature analysis, enabling automated inspection and sorting on the production line.
The camera is fixedly installed above the inspection station, capturing real-time images of the ceramic surface. Through analysis of brightness and saturation channels, it compares changes in glaze reflection and local dark spots, accurately identifying abnormalities such as bubbles, black spots, cracks, and glaze loss.
To ensure accurate and reliable detection results, the system has enhanced the Sobel edge detection and region connectivity analysis algorithms, enabling it to distinguish between glazed decorative patterns and actual crack defects. Under complex lighting conditions, the system maintains recognition stability through automatic exposure and white balance correction, and supports multi-angle sampling calibration to improve detection robustness.
The detection results are transmitted to the main control device via serial port or network interface, controlling the assembly line sorting mechanism to automatically reject defective products.
The actual test results show that this solution achieves an accuracy rate of over 96% in detecting cracks and bubbles at the 0.1 mm level, with an average recognition time of less than 0.8 seconds per item.
SingTown Technology's OpenMV embedded vision system offers advantages such as low cost, high efficiency, high accuracy, and excellent portability. It is particularly suitable for automated inspection projects in research laboratories and online quality inspection systems in small to medium-sized ceramic factories, injecting intelligent detection capabilities into traditional ceramic manufacturing.

AI Sentinel Based on OpenMV: Automatic Alert for Unattended Key Locations
Automatically detects whether the door of a key location remains open for an extended period and issues a timely alert.

Personnel crossing boundary triggers an alarm; OpenMV interprets the “sense of security boundary”
Detecting personnel or equipment crossing virtual perimeter lines.

“AI Urban Management Officer” is Here: An Automated Street Vending Detection System Built on OpenMV
Automatically identifies illegal street vending activities to support daily urban governance.

Improper placement of gas cylinders? Use OpenMV to automatically trigger hazard alerts
Automatically identifies non-compliant placement positions of gas cylinders to proactively detect potential gas safety hazards.

Did you perform live-line work without wearing insulating gloves? OpenMV will immediately issue an alert!
Automatically identifies whether workers are wearing insulating gloves to assist in safe operations.

Utilising the OpenMV smart camera to identify surface defects on aluminium plates in real time
Online identification of defects on aluminium plate surfaces, such as scratches and dents, to support quality control.