Waste Pit Image Analysis for Material Identification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for identifying waste material types in waste treatment plants, such as JP 2007-126246 A and JP 2015-124955 A, face challenges in accurately distinguishing between different types of waste based on color tone and brightness values, leading to potential combustion issues and equipment troubles due to incorrect identification of foreign waste materials.

Innovation Solution

An information processing device and method that uses machine learning to identify waste material types by analyzing image data from the waste pit, employing an identification algorithm that differentiates between various types of waste and non-waste targets, enabling precise control of waste agitation and combustion processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If waste material types are identified by color tone (as in JP 2007-126246 A), then the identification process is simple, but the identification accuracy is insufficient leading to potential combustion issues

Engineering Contradiction:
Improveidentification process complexityVSAvoidwaste material identification accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the identification parameters from simple color tone to multiple parameters including color tone, brightness, and texture features. This allows for more accurate differentiation of waste material types while maintaining a relatively simple image-based identification process.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent adds texture analysis as an additional dimension to the identification process. By combining color tone (2D RGB values) with texture features (spatial frequency domain information), the system achieves better identification accuracy without significantly increasing system complexity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Device complexity

If waste material types are identified by brightness value (as in JP 2015-124955 A), then the detection method is simple, but it cannot accurately distinguish between different types of waste material

Engineering Contradiction:
Improvedetection method complexityVSAvoidwaste material type differentiation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent extends the identification from single-parameter brightness to multi-parameter analysis including brightness, color tone, and texture. This allows the system to differentiate between waste materials with similar brightness values but different textures or colors.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent segments the image analysis into multiple feature extraction stages: color feature extraction, brightness feature extraction, and texture feature extraction. This segmented approach allows each parameter to contribute to the overall identification accuracy while keeping individual processing steps relatively simple.

Inventive Principle:
Principle #1Segmentation

3Measurement precision

If machine learning-based identification is implemented, then waste material identification accuracy is significantly improved, but the device complexity and computational requirements increase

Engineering Contradiction:
Improvewaste material identification accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary feature extraction and preprocessing of image data before feeding it to the machine learning model. By pre-processing the images to extract relevant features (color, brightness, texture) and organizing them in a structured format, the system reduces the computational burden on the machine learning model and speeds up the identification process.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses a pre-trained machine learning model that has been trained on a large dataset of waste material images. This pre-trained model can be deployed as-is or fine-tuned with minimal additional training data, reducing the complexity of model development and deployment while maintaining high identification accuracy.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3842693B1Information processing device, information processing program, and information processing method
Publication Date: 2024.02.28 EBARA ENVIRONMENTAL PLANT
  • EP3842693B1 patent drawingFigure 1
  • EP3842693B1 patent drawingFigure 2
  • EP3842693B1 patent drawingFigure 3

AI summary

Provided is an information processing device capable of identifying a type of waste material in a waste pit. The information processing device includes a type identification unit having an identification algorithm generated by learning training data in which the type of the waste material is labeled on past image data obtained by capturing the inside of the waste pit where the waste material is accumulated, and configured to identify the type of the waste material accumulated in the waste pit when new image data obtained by capturing the inside of the waste pit is given as an input.