Overcoming the Limitations of Legacy Vision Systems with Deep Learning AI
By Shwetha Ramakrishnan, CMO, JIDOKA Technologies Pvt. Ltd.
Bangalore, June 2024 –
In the pursuit of future-fit manufacturing, quality and production teams across industries are focused on achieving "built-in quality." The goal is to ensure that products meet the highest standards from the moment they are created, by proactively addressing quality issues before they escalate into larger problems. This is accomplished by applying continuous improvement tools to optimize the production process without compromising efficiency.
Initially, traditional machine vision solutions were seen as the answer to attaining built-in quality. Production and Quality teams eagerly adopted these systems, which used rules-based algorithms to detect and eliminate objective defects. However, they often missed subjective defects caused by changing conditions on the production line, such as variable lighting, vibrations, and airborne debris. This led to a high incidence of false positives and false negatives, rendering the process suboptimal. Frustrated by the inconsistent performance of these vision solutions, and the high costs and complexity of deployment and integration, manufacturers began bypassing these systems.
This turning point reshaped the industry’s understanding of Quality 4.0. Manufacturers now demand comprehensive vision solutions capable of capturing both objective and subjective defects, both in-line and end-of-line, with 100% system stability under changing line conditions. They also require AI process analytics that enable true process optimization, while minimizing complexity and cost of deployment and integration.
JIDOKA's award-winning Deep Learning Visual Inspection Solutions lead the way in meeting these demands, ensuring "best-in quality" without the risk of high cost of ownership. Our solutions offer:
Pre-Configured Pipelines and Pre-Trained Domain-Specific Models
· Rapid Deployment: Pre-configured use-case pipelines allow for plug-and-play setup with minimal customization.
· Optimized Performance: Pre-trained models adapt to the unique characteristics of your domain, fine-tuning with less training data and time.
Key Benefits: Rapid, low-effort development and deployment, optimized performance.
Unparalleled Configurability and Flexibility
· Adaptive Production Demands and Quality Standards: Designed to cater to changing production demands, crucial for responding to market shifts and disruptions. Accommodates variations and threshold-led nuances for different baseline quality standards without complex parameter tuning, even when these standards differ from customer to customer within the same domain. The adaptive capabilities extend to cost-effective selective checking options, enabling manufacturers to iteratively manage throughput and uphold the highest quality amid evolving production challenges. This flexibility is instrumental in inspecting products at different production stages and specific quality control points, ensuring stringent standards are met for each unique customer requirement.
· User-Friendly UI: Enables non-experts to manage models, defects, thresholds, and acceptance criteria, reducing reliance on specialized resources.
Key Benefits: Optimal quality - throughput trade off, lower cost of ownership, flexibility, highest quality.
Integration with Existing Software and Vision Products
· Hardware Agnostic: Seamlessly integrates with existing vision hardware and software, including ERP, MES Systems, and IIoT.
· Seamless Integration across clouds, on premise/ edge and third party apps.
Key Benefits: Seamless deployment into mass production environments, reduced engineering costs.
Enhanced Product Reliability with 360-Degree Profile Inspection
· Comprehensive Inspection: By stitching together images from multiple surfaces or cameras, we inspect a larger area of the product, increasing defect detection rates.
· Unlimited Defect Varieties: Capable of identifying and detecting unlimited defect varieties without altering part handling.
Key Benefits: Improved product reliability, reduced waste, enhanced customer satisfaction.
Comprehensive Defect Detection Capabilities
· Solutions encompass a wide array of defect detection categories crucial across industries. Surface defects like dents and scratches are identified through advanced AI-driven surface defect detection. Anomaly detection algorithms pinpoint deviations in shape, size, or colour, ensuring adherence to quality standards. Assembly verification checks ensure correct part assembly, while audio-based methods detect variations in decibel levels and vibrations. Identification services verify accuracy in printed materials, including labels and barcodes. Radiography detects internal defects like porosity using X-ray imaging, while print quality inspection ensures precision in color and content. Product inspection covers dimensions and surface defects, while warehouse and logistics inspections verify packaging integrity and counting accuracy.
AI Analytics and Reporting
· Detailed Insights: Provides valuable insights into production performance with detailed analytics and reporting functionalities.
· Proactive Action: Detect defect trends earlier to minimize costly downtimes.
JIDOKA is India’s leading AI visual inspection solutions provider in the Manufacturing Tech space, serving customers across various verticals. Our scalable AI solutions deliver over 12,000 product inspections per minute, consistently maintaining accuracies above 99.5% with a documented 33% increase in throughput and 30-40% reductions in false positives. Digitization returns are realized within 8 -16 months.
In conclusion, JIDOKA's AI Vision Inspection Solutions offer a compelling combination of rapid deployment, flexibility, cost-effectiveness, and data-driven insights that drive measurable business impact. By leveraging JIDOKA's award-winning technology, manufacturers can overcome the limitations of legacy vision systems and achieve built-in quality in their production processes.
Inspired by "Integrated, Centrally-Managed Machine Vision for Built-In Quality" by James Finch, Quality Magazine.
Comments