阅读说明：本技术 检查装置 (Inspection apparatus ) 是由 中谷诚 樽本祥宪 前中章弘 堤弘法 于 2018-10-30 设计创作，主要内容包括：本发明的目的是提供能够抑制包含有时相互重合的多个物品的商品的检查精度降低的检查装置。X射线检查装置(10)对容纳有规定形状的多个物品(A)的商品(G)照射光,基于根据透射过商品(G)的光或者被商品(G)反射的光而得到的检查图像来检查商品(G)。X射线检查装置(10)具备存储部(51)、学习部(52c)以及检查部(52d)。存储部(51)将多个物品(A)重合的状态的商品(G)的检查图像至少存储为指导训练图像。学习部(52c)通过使用存储部(51)中存储的指导训练图像的机械学习来获取关于多个物品(A)重合的状态的商品(G)的特征。检查部(52d)使用学习部(52c)所获取的特征来检查商品(G)。(The invention aims to provide an inspection device capable of preventing the inspection precision of a commodity containing a plurality of articles which are overlapped with each other. An X-ray inspection device (10) irradiates a commodity (G) containing a plurality of articles (A) of a predetermined shape with light, and inspects the commodity (G) based on an inspection image obtained from the light transmitted through the commodity (G) or the light reflected by the commodity (G). An X-ray inspection device (10) is provided with a storage unit (51), a learning unit (52c), and an inspection unit (52 d). A storage unit (51) stores, as at least a guidance training image, an inspection image of a product (G) in a state in which a plurality of articles (A) are stacked. A learning unit (52c) acquires the characteristics of a commodity (G) in a state in which a plurality of articles (A) are superimposed by machine learning using a guide training image stored in a storage unit (51). An inspection unit (52d) inspects the product (G) using the features acquired by the learning unit (52 c).)

1. An inspection device that irradiates a commodity containing a plurality of articles having a predetermined shape with light and inspects the commodity based on an inspection image obtained from the light transmitted through or reflected by the commodity, the inspection device comprising:

a storage unit that stores the inspection image of the product in a state in which the plurality of articles are stacked as at least a guidance training image;

a learning unit that acquires a feature of the commodity with respect to a state in which a plurality of the articles are superimposed, by machine learning using the guidance training image stored in the storage unit; and

an inspection unit that inspects the product using the feature acquired by the learning unit.

2. The inspection apparatus according to claim 1,

the inspection unit inspects whether the product contains foreign matter.

3. The inspection apparatus according to claim 2,

the storage unit stores the inspection image of the commodity including a foreign object as at least the guide training image,

the learning unit acquires the feature by mechanical learning using the guide training image and a region in which a foreign object included in the product is present in the guide training image.

4. The inspection apparatus according to claim 2 or 3,

the storage unit stores, as at least the guide training image, an image obtained by adding a virtual foreign object image to the inspection image of the product not containing a foreign object,

the learning section acquires the feature by mechanical learning using the guide training image and a region in which the virtual foreign object exists in the guide training image.

5. The inspection apparatus according to any one of claims 1 to 4,

the inspection portion inspects the number of the articles contained in the commodity.

6. The inspection apparatus according to claim 5,

the learning section acquires the feature by mechanical learning using the guide training image and the number of the items accommodated by the commodity in the guide training image.

7. The inspection apparatus according to any one of claims 1 to 6,

the inspection unit further inspects the product based on the inspection image without using the feature.

8. The inspection apparatus according to claim 7,

the inspection unit inspects the product based on an inspection result of the product when the feature is used and an inspection result of the product when the feature is not used.