Gradient glass bottles are all subjected to a sandblasting process to create a frosted finish. Common defect types during the manufacturing process are uneven color or black spots on the bottle body. These defects are difficult to detect using the automatic optical inspection (AOI) method because they cannot be clearly defined and their patterns are not fixed.Combining machine vision and artificial intelligence, Solomon uses SolVision to train AI models with glass bottle defect images. By using Segmentation technology to find and learn the feature conditions of defect images, after taking images of the bottle from all angles, the trained AI model can quickly detect the defect distribution of the glass bottle body from all angles and mark the defect locations.