Defect Detection Task Scheduling for Automated Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual defect detection on display screens is slow and prone to human error, leading to unnecessary losses and inefficiencies.
Innovation Solution
A task processing method that includes receiving a detection task, determining its type, and executing it in a preset order using a processor when idle, which involves modifying configuration information, acquiring and training an initial model, and using it for inference to detect defects automatically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual defect detection is performed, then detection accuracy can be maintained through human judgment, but detection speed is slow and productivity is low
Solution Approach 1:
The patent replaces the manual mechanical detection process with an automated deep learning-based image processing system. The processor executes training tasks to build defect detection models and inference tasks to automatically detect defects in display screens, eliminating the need for manual visual inspection while maintaining high detection accuracy.
Solution Approach 2:
The system performs self-training by automatically acquiring training data, preprocessing it, training the deep learning model, and using the trained model for defect detection. The processor manages the entire workflow from model training to defect identification without external intervention, enabling the system to improve its own detection capabilities while maintaining high productivity.
2Reliability
If manual defect detection is performed, then flexibility in handling various defect types is maintained, but human error increases and reliability decreases
Solution Approach 1:
The patent segments the defect detection system into distinct functional modules: data acquisition module, data preprocessing module, model training module, and defect detection module. Each module handles specific aspects of the detection process, making the complex system manageable and reliable. The segmentation allows for specialized processing in each stage, reducing errors while maintaining overall system functionality.
Solution Approach 2:
The system dynamically adjusts training parameters such as learning rate, batch size, and number of epochs based on the specific defect types and product characteristics. Configuration information is modified according to preset rules based on product information, enabling the system to adapt to different detection scenarios while maintaining high reliability through optimized parameters rather than relying on manual judgment consistency.
3Productivity
If automated defect detection is implemented, then productivity increases and detection speed improves, but system complexity and initial setup requirements increase
Solution Approach 1:
The system performs preliminary actions by automatically acquiring training data before defect detection begins. The training data is preprocessed and the deep learning model is trained in advance, so that when actual defect detection is needed, the system can immediately apply the pre-trained model for high-speed automated detection without requiring complex real-time processing setup.
Solution Approach 2:
The patent introduces configuration information as an intermediary that bridges the complex model training process and the simple defect detection operation. The configuration information stores preset rules and parameters that guide the processor in executing training and inference tasks, simplifying the interface between user requirements and complex automated processing while maintaining high productivity.
Data Source
AI summary
The present disclosure relates to a task processing method and device based on defect detection, a computer readable storage medium, and a task processing apparatus. The method includes receiving a detection task; determining a task type of the detection task; storing the detection task in a task queue if the task type is a target task type; and executing the detection task in a preset order and generating a feedback signal when a processor is idle. The detection task of the target task type includes an inference task and a training task. Executing the training task includes modifying configuration information according to a preset rule based on product information in the detection task; acquiring training data and an initial model according to the product information; and using the training data to train the initial model according to the configuration information to obtain a target model and store it in memory.