Scrap determination system and scrap determination method

Patent metadata

Field	Value
Publication number	US-2023173546-A1
Application number	US-202117997368-A
Country	US
Kind code	A1
Filing date	Apr 23, 2021
Priority date	Apr 30, 2020
Publication date	Jun 8, 2023
Grant date	—

Abstract

Provided are a scrap determination system and method that can improve scrap determination techniques. A scrap determination system ( 1 ) comprises: a first scrap determination model ( 221 ) generated using teaching data including first learning images, and determines, based on a camera image, grade of scrap in the image and a ratio of the grade; a second scrap determination model ( 222 ) generated using teaching data including second learning images, and determines, based on the image, grade of scrap in the image and a ratio of the grade; a selection model ( 223 ) configured to determine which of the first scrap determination model ( 221 ) and the second scrap determination model ( 222 ) is to be used, based on the image; and an output section ( 24 ) configured to output information of each grade of scrap and a ratio of the grade determined based on the image using the model selected by the selection model ( 223 ).

First claim

1 . A scrap determination system comprising:
an acquisition section configured to acquire a camera image including scrap; a first scrap determination model generated using teaching data including first learning images, and configured to determine, based on the camera image, each grade of the scrap included in the camera image and a ratio of the grade; a second scrap determination model generated using teaching data including second learning images different from the first learning images, and configured to determine, based on the camera image, each grade of the scrap included in the camera image and a ratio of the grade; a selection model configured to determine which of the first scrap determination model and the second scrap determination model is to be used, based on the camera image; and an output section configured to output information of each grade of the scrap and a ratio of the grade determined based on the camera image using a model selected by the selection model out of the first scrap determination model and the second scrap determination model.

2 . The scrap determination system according to claim 1 , wherein the first learning images are each an image of single-grade iron scrap, and
when determining each grade of the scrap included in the camera image and a ratio of the grade using the first scrap determination model, the ratio of the grade is determined based on an area ratio of scrap of the grade in the camera image.

3 . The scrap determination system according to claim 1 , wherein the second learning images are each an image of mixed-grade iron scrap.

4 . The scrap determination system according to claim 1 , wherein the first learning images, the second learning images, and the camera image are each normalized based on zoom information corresponding to the image.

5 . The scrap determination system according to claim 4 , wherein the first learning images, the second learning images, and the camera image are each normalized to a magnification factor that differs depending on the zoom information corresponding to the image.

6 . The scrap determination system according to claim 1 , wherein at least one of the first scrap determination model, the second scrap determination model, and the selection model is relearned based on the camera image and the information output by the output section.

7 . A scrap determination method that uses: a first scrap determination model generated using teaching data including first learning images, and configured to determine, based on a camera image including scrap, each grade of the scrap included in the camera image and a ratio of the grade; and a second scrap determination model generated using teaching data including second learning images different from the first learning images, and configured to determine, based on the camera image, each grade of the scrap included in the camera image and a ratio of the grade, the scrap determination method comprising:
acquiring the camera image; selecting, based on the camera image, which of the first scrap determination model and the second scrap determination model is to be used, by a selection model; and outputting information of each grade of the scrap and a ratio of the grade determined based on the camera image using a model selected by the selection model out of the first scrap determination model and the second scrap determination model.

8 . The scrap determination system according to claim 2 , wherein the second learning images are each an image of mixed-grade iron scrap.

9 . The scrap determination system according to claim 2 , wherein the first learning images, the second learning images, and the camera image are each normalized based on zoom information corresponding to the image.

10 . The scrap determination system according to claim 9 , wherein the first learning images, the second learning images, and the camera image are each normalized to a magnification factor that differs depending on the zoom information corresponding to the image.

11 . The scrap determination system according to claim 2 , wherein at least one of the first scrap determination model, the second scrap determination model, and the selection model is relearned based on the camera image and the information output by the output section.

Assignees

• Jfe Steel Corp

Inventors

• Kishimoto Yasuo
• Ogasawara Futoshi
• Takao Yasuyoshi

Classifications

• B07C5/342 Primary

  according to optical properties, e.g. colour {(according to radiation transmittivity B07C5/3416)} · CPC title

• G06T7/62

  of area, perimeter, diameter or volume · CPC title

• Y02P10/20

  Recycling · CPC title

• B07C5/3422 Primary

  using video scanning devices, e.g. TV-cameras · CPC title

• Y02W90/00

  Enabling technologies or technologies with a potential or indirect contribution to greenhouse gas [GHG] emissions mitigation · CPC title