Meat Working Point Detection Using Bone-Position Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing working point calculation systems for meat processing face challenges in accurately determining bone positions due to variations in exposure amounts and lack a configuration to reduce the burden of preparing teacher data.

Innovation Solution

A working point calculation system that utilizes a learning model to identify working points in meat by inputting image data to a machine-learned model with simplified teacher data, such as coordinate data, and employs a neural network like HRNet to enhance accuracy and reduce calculation loads.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a learning model is used to obtain the position of the bone part, then the working point can be identified with high accuracy, but the burden of preparing teacher data increases

Engineering Contradiction:

Improveposition identification accuracyVSAvoidteacher data preparation burden

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent uses image data as a copy or representation of the actual meat object, allowing the learning model to process visual information without requiring physical samples or complex physical measurements for teacher data preparation

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces manual or mechanical methods of determining bone positions with a machine learning-based image processing system, where the learning model automatically extracts features and determines positions from images, substituting physical measurement and manual annotation processes

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Device complexity

If conventional methods are used to obtain bone position, then the system is simpler to implement, but the position accuracy decreases due to exposure variation

Engineering Contradiction:

Improvesystem implementation simplicityVSAvoidbone position accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the parameters used for bone position detection from raw pixel intensity values (which are affected by exposure) to extracted features such as contours, edges, and shape characteristics that are more invariant to lighting conditions and exposure variations

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary processing stage where image features are extracted and processed before final position determination. This intermediary layer (feature extraction and learning model processing) acts as a mediator that transforms raw image data into robust position information that is less sensitive to exposure variations

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentEP4316246B1Action point calculation system for meat, meat processing system, and action point calculation method for meat

Publication Date: 2025.10.29 MAYEKAWA MFG CO LTD

Figure 1
Figure 2
Figure 3

AI summary

A working point calculation system for meat, includes: an image data acquisition unit; and a working point acquisition unit. The image data acquisition unit is configured to acquire image data indicating a photographed image of meat. The working point acquisition unit is configured to acquire working point data for identifying at least one working point where a robot gives working in the meat, by inputting the image data acquired by the image data acquisition unit to a learning model machine-learned using, as teacher data, the image data and correct answer data indicating a key point of the meat included in the photographed image.