Inspection device, inspection method, and storage medium storing program

The invention provides an inspection device for inspecting a substrate having a pattern on the surface. The inspection device is provided with a classification unit (303) and an output unit (304). A classification unit (303) uses a learned model created by machine learning to classify defect candidates detected by a single inspection performed on an image to be inspected obtained by capturing an image of a substrate for each type of defects and false alarms. The output unit (304) outputs the classification result by the classification unit (303). As a result, it is possible to feed back the classification results of defects and false alarms to a previous step (for example, a substrate manufacturing step, a primary inspection step, or the like). 本发明提供一种对在表面上具有图案的基板进行检查的检查装置，该检查装置具有分类部(303)和输出部(304)。分类部(303)利用通过机器学习创建的已学习模型，按照缺陷以及虚报的种类对由针对拍摄基板而得到的被检查图像进行的一次检查而检测出的缺陷候补进行分类。输出部(304)输出基于分类部(303)的分类结果。由此，能够将缺陷以及虚报的分类结果反馈给前工序(例如，基板的制造工序、一次检查工序等)。