Camera-Based Scrap Grading With Image Segmentation and AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scrap metal discrimination technologies face challenges in automating the grading process due to varying scrap shapes and sizes, leading to inconsistent human-based visual discrimination and inefficiencies, particularly in cases where cranes cannot be used.
Innovation Solution
A scrap discrimination system utilizing machine learning models to extract and grade scrap parts from camera images, employing semantic segmentation and neural networks to identify and classify scrap grades and foreign objects, eliminating the need for human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual discrimination by workers is used, then scrap grade can be determined, but inconsistency in discrimination results occurs due to worker skill level and aging
Solution Approach 1:
The patent replaces the mechanical visual inspection process performed by workers with an automated image processing system using cameras and algorithms. The system captures images of scrap metal and automatically determines grades based on shape, size, and other visual characteristics, eliminating human skill level variations and ensuring consistent discrimination results.
Solution Approach 2:
The system enables self-service automation where the scrap grading process no longer requires human operators. The automated system independently performs image capture, analysis, and grade determination, making the process self-sufficient and eliminating dependence on worker availability and skill levels.
2Measurement precision
If sequential determination while suspending scrap with magnetic crane is used, then scrap grade can be measured, but the process takes long time
Solution Approach 1:
The patent replaces the mechanical sequential determination process with a parallel automated image processing system. Multiple scrap pieces can be captured and analyzed simultaneously by the camera system, and the processing algorithm evaluates all pieces in parallel, dramatically reducing the time required compared to sequential magnetic crane-based determination.
Solution Approach 2:
The system performs preliminary image capture of multiple scrap pieces before analysis. By capturing all necessary images in advance and then processing them through automated algorithms, the system enables rapid batch determination rather than sequential one-by-one analysis, improving overall productivity.
3Measurement precision
If automated image processing is implemented, then discrimination consistency is improved, but complexity of extracting accurate scrap features from varied shapes and sizes increases
Solution Approach 1:
The patent applies segmentation by dividing the complex scrap recognition task into distinct processing stages: image capture, preprocessing, feature extraction, and grade determination. Each stage handles specific aspects of the problem, making the overall complex process more manageable and effective.
Solution Approach 2:
The system transforms the complex problem of recognizing varied scrap shapes and sizes by changing the parameters used for analysis. Instead of attempting to recognize complete complex shapes, the system extracts key parameters such as area, perimeter, aspect ratio, and other simplified geometric features that effectively characterize scrap for grading purposes.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A scrap discrimination system and a scrap discrimination method that can improve scrap discrimination technology are provided. A scrap discrimination system includes a scrap part extraction model (221) that extracts, based on a camera image, a scrap part located in a central portion included in the camera image with reference to a window (107) defined in advance in an image, a scrap discrimination model (222), generated by teacher data including training images, that sorts grades of scrap and a ratio of each grade from a scrap image extracted by the scrap part extraction model (221), and an output interface (24) that outputs information on the grades of scrap and the ratio of each grade as discriminated based on the scrap image using the scrap discrimination model (222).