Why This Algorithm Was Chosen
• High Accuracy → 94.5%+ recognition rate using pretrained vision-language models
• Zero Training No datasat coollection or model training required
• Universal Support Recognizes any global currency without retraining
• Robust Detection Handles damaged notes, poor lighting, and varied orientations
• No Maintenance Cloud infrastructure automatically updated by Google
• Lightweight Deployment No local model storage (saves 4.2MB vs CNN)
• Multimodal Understanding – Reads text, symbols, and visual patterns simultaneously
