Search

CN-121999191-A - Image storage device for learning and commodity calling device

CN121999191ACN 121999191 ACN121999191 ACN 121999191ACN-121999191-A

Abstract

The invention provides a learning image storage device and a commodity calling device, wherein the learning image storage device stores learning images for learning images related to articles. The learning image storage device includes an imaging unit that images an article, a determination unit that determines whether or not an imaged image of the imaged article is an abnormal image, and a storage unit that stores at least a part of the imaged image as a learning image. The determination unit determines whether or not the captured image is an abnormal image based on the learned model related to the article and the feature amount of the captured image, and the storage unit stores, as the learning image, the captured image that is determined not to be the abnormal image among the plurality of captured images.

Inventors

  • Tateishi kyota
  • Yasukuni onohara
  • IWAKAWA Ken

Assignees

  • 株式会社石田

Dates

Publication Date
20260508
Application Date
20251106
Priority Date
20241107

Claims (4)

  1. 1. An image storage device for learning, which stores a learning image for learning an image related to an article, The learning image storage device includes: A photographing unit that photographs the object; A determination unit for determining whether or not the captured image of the object is an abnormal image, and A storage unit that stores at least a part of the captured image as the learning image, The determination unit determines whether or not the captured image is the abnormal image based on a learned model related to the article and the feature amount of the captured image, The storage unit stores, as the learning image, the captured image determined not to be the abnormal image among the plurality of captured images.
  2. 2. The learning image storage device of claim 1, wherein, The determination unit determines that the captured image is the abnormal image when the degree of abnormality calculated based on the learned model and the feature of the captured image is equal to or greater than a predetermined abnormality determination threshold, The later the shooting period of the shot image is, the smaller the degree of abnormality is calculated.
  3. 3. The learning image storage device according to claim 1 or 2, wherein, The determination unit determines whether the captured image is the abnormal image based on a result of comparing the degree of abnormality calculated based on the learned model and the feature of the captured image with a predetermined abnormality determination threshold, The abnormality determination threshold is a value of the degree of abnormality in which a predetermined proportion of the captured images are determined to be abnormal.
  4. 4. A commodity calling device is characterized by comprising: The learning image storage device according to claim 1 or 2; An imaging unit that images a commodity; a commodity information storage unit for storing commodity information related to the type of the commodity, and A calling unit that calls the commodity information, The calling unit determines the type of the commodity by applying the feature amount of the captured image of the commodity to the model after learning using the learning image, and calls the commodity information corresponding to the determined type of the commodity.

Description

Image storage device for learning and commodity calling device Technical Field The present invention relates to a learning image storage device and a commodity calling device. Background Patent document 1 describes a technique for automatically collecting training data concerning an article by storing an image of the article in association with article information when a new article is conveyed by a conveying section and when characteristics of the article change and cannot be handled by a previously constructed learning model in an article processing apparatus. Prior art literature Patent literature Patent document 1 Japanese patent No. 7368834 Disclosure of Invention Problems to be solved by the invention As in the prior art described above, when automatically collecting learning images (e.g., training data) related to an article, it is desirable to remove abnormal images including noise from a large number of images including the article. However, when the number of images including the article is large, it is not easy for the operator to determine whether or not the image is an abnormal image. The present invention aims to provide a learning image storage device and a commodity calling device, which can store a shooting image which is obtained by automatically removing an abnormal image containing noise from a plurality of shooting images as a learning image. Means for solving the problems (1) A learning image storage device for storing a learning image for learning an image related to an article includes an imaging unit for imaging the article, a determination unit for determining whether or not the imaged image of the imaged article is an abnormal image, and a storage unit for storing at least a part of the imaged image as the learning image, wherein the determination unit determines whether or not the imaged image is an abnormal image based on a learning model related to the article and a feature amount of the imaged image, which are previously constructed, and the storage unit stores, as the learning image, the imaged image determined not to be the abnormal image among the plurality of imaged images. In the learning image storage device according to one embodiment of the present invention, the determination unit determines whether or not the captured image is an abnormal image based on the learned model related to the article and the feature amount of the captured image. The storage unit stores, as a learning image, a captured image that is determined not to be an abnormal image among the plurality of captured images. Thus, for example, every time an object is photographed by the photographing section, photographed images determined not to be abnormal images are automatically sequentially stored as learning images. Therefore, according to the learning image storage device of the embodiment of the present invention, it is possible to store, as the learning image, the captured image from which the abnormal image including noise is automatically removed from the plurality of captured images. (2) In the above (1), the determination unit may determine that the captured image is an abnormal image when the degree of abnormality calculated based on the learned model and the feature of the captured image is equal to or greater than a predetermined abnormality determination threshold, and may calculate the degree of abnormality to be smaller as the captured image is captured later. In this case, since it is more difficult to determine that the captured image is an abnormal image as the captured image is captured later, for example, when only a part of the articles is changed to the latest standard, it is possible to suppress erroneous determination of the captured image that does not actually include noise as an abnormal image. (3) In the above (1) or (2), the determination unit may determine whether or not the captured image is an abnormal image based on a comparison result between the degree of abnormality calculated based on the learned model and the feature of the captured image and a predetermined abnormality determination threshold, the abnormality determination threshold being a value of degree of abnormality in which a predetermined proportion of the captured images in the captured image are determined to be abnormal. In this case, for example, if the occurrence rate of an abnormality including noise in a plurality of captured images is known in advance, the abnormality determination threshold value can be automatically set. (4) The commodity calling device according to another aspect of the present invention may include the learning image storage device according to (1) or (2), a shooting unit that shoots a commodity, a commodity information storage unit that stores commodity information related to a type of the commodity, and a calling unit that calls the commodity information, wherein the calling unit determines the type of the commodity by applying a feature value of a shot image of the shot commo