AI-POWERED DEFECTS DETECTION SYSTEM FOR ALLOY WHEEL MANUFACTURE

POC ALIGNED WITH GM
STANDARDS
PILOT EXECUTION ON A
SINGLE LINE
SCALE-UP ACROSS ALL
PRODUCTION LINES
DEPLOYMENT OF
INDUSTRIAL HIGH
RESOLUTION CAMERAS
CONTROLLED LIGHTING FOR
CONSISTENT IMAGING
USE OF CNN AND SEMANTIC
SEGMENTATION MODELS
REAL-TIME PROCESSING
WITH EDGE COMPUTING
AUTOMATED DEFECT
DETECTION AND
CATEGORIZATION
TRIGGERED REJECTION FOR
IDENTIFIED DEFECTS

Transcript

1. AI-POWERED DEFECTS DETECTION SYSTEM FOR ALLOY WHEEL MANUFACTURER
2. MANUAL INSPECTION PROCESSES IN ALLOY WHEEL MANUFACTURING WERE INCONSISTENT, SLOW, AND ERROR-PRONE, LEADING TO HIGH DEFECT RATES AND COMPROMISED QUALITY. PROBLEM STATEMENT AND CHALLENGES
3. SOLUTION APPROACH POC ALIGNED WITH GM STANDARDS PILOT EXECUTION ON A SINGLE LINE SCALE-UP ACROSS ALL PRODUCTION LINES DEPLOYMENT OF INDUSTRIAL HIGH- RESOLUTION CAMERAS CONTROLLED LIGHTING FOR CONSISTENT IMAGING USE OF CNN AND SEMANTIC SEGMENTATION MODELS REAL-TIME PROCESSING WITH EDGE COMPUTING AUTOMATED DEFECT DETECTION AND CATEGORIZATION TRIGGERED REJECTION FOR IDENTIFIED DEFECTS
4. VALUE CREATION Quantified Benefits in 6 Months: 95%+ defect detection accuracy; 70% drop in false negatives. 90% reduction in manual inspection labor. 30% increase in production throughput. 25% drop in returns and rework. Enhanced traceability and audit readiness for OEMs. ROI Projection: Final ROI will be determined within 9–12 months, factoring savings from labor, reduced scrap, and compliance improvements.
5. TECHNOLOGIES USED GOAL 1 GOAL 1 AI & Computer Vision Stack: CV & ML Libraries: OpenCV, YOLOv8, TensorFlow, PyTorch Model Types: Custom CNNs, Semantic Segmentation, Transfer Learning Hardware & Infrastructure: 4K Industrial Cameras, GPenabled Edge Devices (NVIDIA Jetson) Lighting System: 1000 ± 100 Lux Annotation: CVAT, Labelbox Integration: MES, Secure Encrypted Data Flow & Access Controls
6. SCAN HERE TO SEE A DEMO ON YOUR DATASETS THANK YOU