SolVisionCase Study
Metal Bolt Defect Detection Using AI
Customer
The customer is a leading nuts and bolts manufacturer headquartered in Asia.
Case
Metal Bolt Quality Control
The customer sought a precise defect detection solution capable of identifying defects, such as small dents, on manufactured bolts, irrespective of the position or orientation of the bolts.
Challenge
Reducing Defect Detection False Negative Rate
To prevent redundancy in defect detection, each bolt needed to be analyzed individually from three different angles. Due to the large production volume, the customer required the inspection process to be completed within 1 second per bolt, while maintaining a minimum accuracy rate of 90%. This task was further complicated by the reflective surface of the bolts, as light reflections could be misinterpreted as defects.
Solution
Innovative AI Inspection Solution
Solomon devised a robot-assisted system that turns each bolt 360 degrees, capturing three images to identify any defects. To address reflective surface challenges, an AI model in SolVision was trained, and our engineers provided the customer with recommendations for optimal lighting and background conditions to minimize reflections. This integrated approach enabled a rapid inspection process, completing inspection in just 49 milliseconds per bolt with a flawless 100% accuracy rate.