AI-Powered Visual Inspection in Konark Fixtures
Powered by AeonX Solutions
About the Customer
KONARK GROUP is having a turnover of above 200 crores ($50 million). Its 11 manufacturing units are located in Daman, Gujarat, Maharashtra & HP, having floor space area of approx. 20,000 sq. meters & more. They are into Herbal Extracts & Health Care, Flavors & Fragrances, Domestic UV & RO Water Purification System, Water & Wastewater Treatment Plants, Rainwater Harvesting, Testing & Calibration Laboratory, Designing & Testing Centre Bioremediation Solutions, Luminaries. The first luminaries manufacturing unit of the group was established in 1972 and over the years it has progressed to be one of the leading manufacturers of HID Luminaries, specializing in road lighting and outdoor luminaries & accessories. It is their belief that good growth can be achieved only through proven quality system. Konark’s journey towards planned quality systems began in 1995, initiating actions to obtain ISO Certification. Konark is also a recipient of Ramkrishna Bajaj National Quality Award (R.B.N.Q.A.) for 2003 – 2004 which is equivalent to Malcolm Award, U.S.A.
The Challenge
Konark Fixtures Limited, a leading automotive components manufacturer, was facing escalating quality assurance issues on its assembly lines. Their existing manual inspection processes were increasingly inadequate in the face of growing production volumes and customer quality expectations.
Business Challenges
Inconsistent quality control
Human inspectors often missed subtle defects, leading to increased defect rates and rework.
High labor costs
Maintaining 100% manual inspections required significant workforce allocation, impacting profitability.
Customer dissatisfaction and warranty claims
Defects slipping through QA processes led to returns and reputational damage.
Scalability limitations
Manual inspection couldn’t scale with rising throughput and variant complexity across product lines.
Technical Challenges
No real-time inspection feedback
Quality issues were often discovered post-production, delaying corrective actions.
Limited traceability and defect logging
Existing systems lacked structured defect logging or digital audit trails for analytics.
Non-standardized inspection parameters
Inspection criteria varied between operators, leading to inconsistent evaluation.
No predictive quality analytics
Lack of integration with ML models or structured data to anticipate defect trends or process drifts.
Inefficient alerting system
Notifications of anomalies were delayed or not actionable, reducing responsiveness on the shop floor.
Risks and Impact if the Challenge Were Not Addressed
Failure to modernize Konark’s inspection and quality control process would have resulted in significant operational, financial, and reputational risks:
Business Risks
Escalating defect rates
Continued reliance on manual inspection could lead to undetected defects, increasing product rework, customer complaints, and warranty claims.
Operational inefficiencies
High dependency on human inspectors would have strained scalability, especially with the launch of new product variants and higher production volumes.
Cost overrun
Increased labor and material costs due to rejections and rework would reduce profitability and operational margins.
Customer attrition
Persistent quality issues would jeopardize relationships with key OEM clients, risking revenue loss and long-term contracts.
Technical Risks
Lack of traceability and auditability
Without structured digital inspection logs, Konark would be unable to perform root cause analysis or meet compliance requirements.
No real-time defect resolution
Inability to flag defects as they occur would result in batch-level quality failures instead of single-instance corrections.
Data silos and manual reporting
Absence of a centralized defect data repository and analytics would prevent continuous improvement or process optimization.
Missed opportunity for automation
Without AI/ML-based quality checks, the organization would fall behind in adopting Industry practices, impacting competitiveness in a tech-driven market.
Ultimately, the absence of a robust AI-driven inspection system would hinder Konark’s transformation into a smart manufacturing enterprise, affecting both short-term operational excellence and long-term strategic goals.
Partner Solution
To address Konark Fixtures Limited’s quality control challenges, AeonX Digital implemented a robust Generative AI-powered visual inspection system leveraging AWS capabilities. The solution combined computer vision-based defect detection with natural language generation, enabling automated, real-time quality checks that drastically reduced human error and inspection delays.
01
Key highlights of the implementation using YOLOv11:
- Deployed custom-trained YOLOv11 object detection models tailored to each product family
- Leveraged transfer learning from pre-trained industrial models to accelerate model training
- Enabled real-time inference using GPU-accelerated edge devices for high-speed image processing
- Detected:
- Component presence or absence
- Correct positioning and alignment
- Assembly sequence accuracy
- Surface-level defects and anomalies
02
Amazon Bedrock integration and GenAI capabilities
- Used Amazon Bedrock to translate technical detections into business-friendly insights
- Engineered prompts for:
- Natural language summaries of detected quality issues
- Structured grading of component quality with justifications
- Technical explanations for anomalies detected on the line
- Trend analysis to identify recurring or emerging quality patterns
- Enabled actionable outputs such as:
- Recommended corrective actions
- Quality classification tags
- Historical context referencing similar past issues
03
Workflow automation and system integration
- Developed a complete end-to-end workflow:
- High-resolution CCTV images captured per unit
- YOLOv11 models triggered to detect and classify issues
- Bedrock-generated insights pushed to dashboards and alerting systems
- Integrated with Konark’s existing Quality Management System (QMS) for seamless defect tracking
- Configured real-time alerts via WhatsApp, email, and web dashboards
- Implemented:
- Secure image storage with automated retention policies
- Multi-camera coordination for 360° inspection coverage
- Custom visualization dashboards for inspection performance and metrics
Accelerate Inspection. Improve Accuracy.
Real-time AI detection is helping manufacturers stay ahead of defects and delays.
