Among the five modes, the Classification, Detection, and Segmentation modes are classified as the Defect Inspection mode. The Defect Inspection mode basically determines whether an image is defective or not, and is classified as follows according to detailed features. The **Classification** mode determines whether each image is OK or defective, and, if it is defective, it determines what kind of defect it is. The **Detection** mode detects a defective area in each image as a rectangle, while the **Segmentation** mode detects a defective area in pixels.
The Defect Inspection mode is largely composed of the Developer project that train a deep learning model through the image and defect information of the product entered by users, and the Inspector project that inspect new images for defects using the training results acquired in the Developer. A Developer project consists of three tabs: [Database], [Train], and [Analyze], and an Inspector project consists of one tab: [Inspect]. Users can create multiple subtabs in the [Inspect] tab of the Inspector project, and perform defect inspections using different training results.