Inspection device and inspection method
An inspection device includes: a first acquisition unit for acquiring a first defect probability calculated using a first learning model from data regarding an inspection target processed in a first manufacturing process; a second acquisition unit for acquiring a second defect probability calculated using a second learning model from data regarding the inspection target processed in a second manufacturing process after the first manufacturing process; and a determination unit for determining a defect in the inspection target if the acquired second defect probability is greater than or equal to a second threshold value, the determination unit being configured to change at least one of the second learning model and the second threshold value into a condition for enhancing inspection accuracy if the acquired first defect probability is greater than a predetermined first threshold, compared to if the first defect probability is less than or equal to the first threshold value.
1. An inspection device comprising a processor and a memory, the processor executing instructions stored in the memory and configured to:
acquire a first defect probability, the first defect probability being calculated using a first learning model into which data regarding an inspection target processed in a first manufacturing process is input; and
acquire a second defect probability, the second defect probability being calculated using a second learning model into which data regarding the inspection target processed in a second manufacturing process performed after the first manufacturing process is input, wherein
(a) when the acquired first defect probability is less than or equal to a first threshold value, the processor determines that the inspection target has a defect when the acquired second defect probability is greater than or equal to a second threshold value, the second defect probability being calculated using a first model of the second learning model,
(b) when the acquired first defect probability is greater than the first threshold value:
(b1) the processor determines that the inspection target has a defect when the acquired second defect probability is greater than or equal to the second threshold value, the second defect probability being calculated using the first model of the second learning model,
(b2) when the acquired second defect probability that was calculated using the first model of the second learning model is less than the second threshold value, the processor re-calculates the second defect probability using a second model of the second learning model, the second model being different from the first model, the second model having a higher accuracy than the first model, the first model having a higher processing speed than the second model, and
(b3) the processor determines that the inspection target has a defect when the acquired second defect probability re-calculated using the second model is greater than or equal to the second threshold value.
2. The inspection device according to claim 1 , wherein the first model is a Convolutional Neural Network (CNN), and the second model is a region-based CNN (R-CNN).
3. An inspection method performed by a processor executing instructions stored in a memory, the method comprising:
acquiring a first defect probability, the first defect probability being calculated using a first learning model into which data regarding an inspection target processed in a first manufacturing process is input; and
acquiring a second defect probability, the second defect probability being calculated using a second learning model into which data regarding the inspection target processed in a second manufacturing process performed after the first manufacturing process is input, wherein
(a) when the acquired first defect probability is less than or equal to a first threshold value, the processor determines that the inspection target has a defect when the acquired second defect probability is greater than or equal to a second threshold value, the second defect probability being calculated using a first model of the second learning model,
(b) when the acquired first defect probability is greater than the first threshold value:
(b1) the processor determines that the inspection target has a defect when the acquired second defect probability is greater than or equal to the second threshold value, the second defect probability being calculated using the first model of the second learning model,
(b2) when the acquired second defect probability that was calculated using the first model of the second learning model is less than the second threshold value, the processor re-calculates the second defect probability using a second model of the second learning model, the second model being different from the first model, the second model having a higher accuracy than the first model, the first model having a higher processing speed than the second model, and
(b3) the processor determines that the inspection target has a defect when the acquired second defect probability re-calculated using the second model is greater than or equal to the second threshold value.
4. The inspection method according to claim 3 , wherein the first model is a Convolutional Neural Network (CNN), and the second model is a region-based CNN (R-CNN).