X-ray image processing system and X-ray image processing method

Abstract

To provide an X-ray image processing system or the like for appropriately discriminating an article based on an X-ray image.SOLUTION: An X-ray image processing system 100 includes a control unit 11 for discriminating a category of an article based on an X-ray image of the article. The control unit 11 changes a recognition category of the article to a predetermined candidate category based on the comparison of the recognition category based on the image recognition of the article and a plurality of candidate categories based on the operation of a user, and displays the candidate category on a display device in association with the X-ray image of the article.SELECTED DRAWING: Figure 2

Claims

[Claim 1] The system includes a control unit that determines the category of an item based on an X-ray image of the item, The control unit, based on a comparison between the recognition category of the item based on image recognition and a plurality of candidate categories based on user operation, changes the recognition category of the item to a predetermined candidate category, and displays the candidate category on the display device in association with the X-ray image of the item. An item recognition unit that determines the category of an item on the X-ray image based on the item recognition model that performs the aforementioned image recognition, A change history table that shows the correspondence between the original category of an item assigned by user input and the changed category which looks similar to the original category, The system further comprises a recognition category determination unit that, if the determination result of the item recognition unit corresponds to the changed category, determines the category of the item based on the recognition category determination model, The aforementioned recognition category determination model is a model for determining whether an item that looks similar to an item in the changed category belongs to the category before the change. The aforementioned item recognition model learns that for items that look similar to the modified category but are originally in the pre-modification category, the correct category for such items is set to the modified category. The control unit, as part of the comparison between the recognized category and the candidate category, determines, based on the determination result of the recognized category determination unit, that the candidate category, which is the category of the article before the change, has a higher reliability than the recognized category, which is the category of the article after the change, and then sets the category of the article to the candidate category and displays the candidate category on the display device in association with the X-ray image of the article, in an X-ray image processing system. [Claim 2] When the control unit changes the recognition category of an item to a predetermined candidate category, it displays a predetermined character indicating the candidate category of the item and another character indicating the recognition category of the item in association with the X-ray image of the item. The X-ray image processing system according to claim 1, characterized by the following: [Claim 3] Based on user input, if a predetermined candidate category is assigned to an item in order to correct misrecognition or omission of the item's category, the control unit registers the X-ray image of the item as additional training data. The control unit, when the confidence level of image recognition of the item in additional training is above a predetermined value, and the recognition category of the item is different from the candidate category, changes the category of the item from the candidate category to the recognition category and then generates the item recognition model to be used for image recognition of the item. The X-ray image processing system according to claim 1, characterized by the following: [Claim 4] When the control unit determines the category of an item based on its X-ray image, and the misrecognition or omission of the item's category is not corrected by user input, it associates the X-ray image of the item with the item's category and registers it as additional training data. The X-ray image processing system according to claim 3, characterized by the following: [Claim 5] When the control unit performs additional training on the object recognition model used for image recognition of objects, it generates the change history table that shows the history of changing the category of an object on an X-ray image from a predetermined candidate category to the recognition category. The X-ray image processing system according to claim 1, characterized by the following: [Claim 6] When the category of an item is changed from the recognized category to the candidate category, the control unit associates the recognized category with the X-ray image of the item when using the X-ray image of the item in the next additional training of the item recognition model. The X-ray image processing system according to claim 5, characterized by the following: [Claim 7] When the control unit generates the recognition category determination model during additional learning, for items where the recognition category and the candidate category are different, it associates the candidate category with the X-ray image of the item. The X-ray image processing system according to claim 5, characterized by the following: [Claim 8] The control unit, after performing image recognition of an item using the recognition category determination model, determines that the confidence level of the item's category when it is a candidate category is less than or equal to the confidence level of the item when it is a recognized category, and furthermore, if the categories ranked second or lower in the results of image recognition of the item using the item recognition model correspond to predetermined priority categories, then changes the item's category from the recognized category to the priority category and displays the priority category on the display device in association with the X-ray image of the item. The X-ray image processing system according to claim 5, characterized by the following: [Claim 9] When the category of an item is changed from the recognition category to the preferred category, the control unit associates the recognition category with the X-ray image of the item when using the X-ray image of the item in the next additional training of the item recognition model. The X-ray image processing system according to claim 8, characterized by the following: [Claim 10] Neither the preferred category included in the predetermined preferred category table nor the predetermined category before the change to the preferred category overlaps with the recognized category included in the change history table. The X-ray image processing system according to claim 8, characterized by the following: [Claim 11] The control unit, if the second-ranked or lower-ranked categories in the image recognition results of an item correspond to a predetermined priority category included in the candidate categories, changes the category of the item to the priority category and displays the priority category on the display device in association with the X-ray image of the item. The X-ray image processing system according to claim 1, characterized by the following: [Claim 12] A discrimination process that determines the category of an item based on an X-ray image of the item, The system includes a display process that, based on a comparison between a recognition category based on image recognition of an item and a plurality of candidate categories based on user operation, changes the recognition category of the item to a predetermined candidate category and displays the candidate category on a display device in association with the X-ray image of the item, The aforementioned discrimination process includes: Based on the aforementioned image recognition object recognition model, an object recognition process is performed to determine the category of the object on the X-ray image, The process includes a recognition category determination process that, based on a change history table showing the correspondence between the original category of an item assigned by user input and the modified category that looks similar to the original category, determines the category of the item based on a recognition category determination model if the determination result of the item recognition process corresponds to the modified category. The aforementioned recognition category determination model is a model for determining whether an item that looks similar to an item in the changed category belongs to the category before the change. The aforementioned item recognition model learns that for items that look similar to the modified category but are originally in the pre-modification category, the correct category for such items is set to the modified category. In the comparison between the recognized category and the candidate category, if the determination result of the recognized category determination process indicates that the candidate category, which is the category of the item before the change, has a higher reliability than the recognized category, which is the category of the item after the change, then in the display process, the category of the item is changed to the candidate category, and the candidate category is associated with the X-ray image of the item and displayed on the display device, an X-ray image processing method.

Images