Information processing device and learning recognition system

Patent metadata

Field	Value
Publication number	US-12333409-B2
Application number	US-202117513968-A
Country	US
Kind code	B2
Filing date	Oct 29, 2021
Priority date	Jan 13, 2021
Publication date	Jun 17, 2025
Grant date	Jun 17, 2025

Abstract

An information processing device includes a parallel deep neural network configured to input a captured image of an article to deep neural network models respectively corresponding to a plurality of articles and perform inferences about the plurality of articles in parallel using the deep neural network models, a new article determination unit configured to determine whether an article included in the image is an unlearned article based on learned model information about the articles and the image, and a new article learning unit configured to learn a deep neural network model corresponding to the article determined to be unlearned based on the image and initial model configuration information about the deep neural network model when the article included in the image is determined to be an unlearned article. The new article learning unit adds the learned deep neural network model to the deep neural network models.

First claim

What is claimed is:

1. An information processing device, comprising:
a parallel deep neural network configured to input a captured image of an article to deep neural network models respectively corresponding to a plurality of articles and perform inferences about the plurality of articles in parallel using the deep neural network models;
a new article determination unit configured to determine whether an article included in the image is an unlearned article based on learned model information about the articles and the image; and
a new article learning unit configured to learn a deep neural network model corresponding to the article determined to be unlearned based on the image and initial model configuration information about the deep neural network model when the article included in the image is determined to be an unlearned article; and
a shape detection unit configured to detect a shape of the article from the image,
wherein the new article learning unit adds the learned deep neural network model to the deep neural network models,
wherein the new article learning unit learns the deep neural network model corresponding to the article included in the image based on configuration information indicating a feature corresponding to the detected shape of the article, and
wherein the parallel deep neural network includes:
a feature extraction unit configured to extract a feature from the image, and
an identification unit configured to output a confidence value indicating a probability that an article corresponding to the deep neural network model exists based on the extracted feature, and

wherein the feature extraction unit is shared for each detected shape of the article.

2. The information processing device according to claim 1 ,
wherein the parallel deep neural network uses the deep neural network models including the added learned deep neural network model to perform inferences about a plurality of articles including the article determined to be unlearned in parallel.

3. The information processing device according to claim 1 ,
wherein the new article learning unit performs the learning without setting a part of parameters of the deep neural network model corresponding to the article determined to be unlearned.

4. The information processing device according to claim 1 ,
wherein the parallel deep neural network includes:
a feature extraction unit configured to extract a feature from the image, and
an identification unit configured to output a confidence value indicating a probability that an article corresponding to the deep neural network model exists based on the extracted feature, and

wherein the feature extraction unit is shared for the plurality of articles.

5. The information processing device according to claim 1 ,
wherein the parallel deep neural network performs the inferences by preferentially using a deep network model corresponding to an article with a shape the same as that of a previously detected article.

6. The information processing device according to claim 1 , further comprising:
a multi-article deep neural network that is constituted by a learned deep neural network model for a predetermined type of article,
wherein the parallel deep neural network and the multi-article deep neural network perform inferences for the plurality of articles in parallel.

7. A learning recognition system, comprising:
a camera configured to capture an image of an article;
a transport device configured to transport an article from an input side to an output side;
a control device configured to control grasping of an article transported by the transport device; and
an information processing device configured to perform learning using a deep neural network,
wherein the information processing device includes:
a parallel deep neural network configured to input an image of an article captured by the camera to deep neural network models respectively corresponding to a plurality of articles and perform inferences about the plurality of articles in parallel using the deep neural network models,
a new article determination unit configured to determine whether an article included in the image is an unlearned article based on learned model information about the articles and the image,
a new article learning unit configured to learn a deep neural network model corresponding to the article determined to be unlearned based on the image and initial model configuration information about the deep neural network model when the article included in the image is determined to be an unlearned article, and
a shape detection unit configured to detect a shape of the article from the image,

wherein the new article learning unit adds the learned deep neural network model to the deep neural network models,
wherein the parallel deep neural network outputs a recognition result of the article included in the image by using deep neural network models including the added learned deep neural network model,
wherein the control device controls grasping of the article based on the recognition result,
wherein the new article learning unit learns the deep neural network model corresponding to the article included in the image based on configuration information indicating a feature corresponding to the detected shape of the article,
wherein the parallel deep neural network includes:
a feature extraction unit configured to extract a feature from the image, and
an identification unit configured to output a confidence value indicating a probability that an article corresponding to the deep neural network model exists based on the extracted feature, and

wherein the feature extraction unit is shared for each detected shape of the article.

8. The learning recognition system according to claim 7 ,
wherein the control device controls grasping of an article transported by the transport device having a loop shape.

Assignees

• Hitachi Ltd

Inventors

• Oshima Takashi
• Ono Goichi
• Kishimoto Tadashi
• Kokubo Masaru

Classifications

• G06N3/0495

  Quantised networks; Sparse networks; Compressed networks · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/082

  modifying the architecture, e.g. adding, deleting or silencing nodes or connections · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06V10/82

  using neural networks · CPC title