Back to Projects
Computer Vision· 2025
Saudi Currency Recognition
Real-time computer vision system for currency identification
6
Denominations
1,330
Reference Images
5
Max Notes/Frame
5,000
SIFT Keypoints
Problem
Manual currency counting is slow and error-prone. Businesses handling cash transactions need automated denomination identification and authenticity verification.
Solution
Built a real-time currency recognition system using SIFT feature extraction and FLANN-based matching that identifies Saudi Riyal denominations from live camera feeds with authenticity verification.
Tech Stack
OpenCVSIFTFLANNNumPyStreamlitPandasPython
Key Features
Live camera recognition with real-time denomination identification
Multi-note detection (up to 5 notes per capture) using Canny edge detection
SIFT feature extraction with FLANN-based nearest neighbor matching
Authenticity verification using color deviation analysis
Currency aggregation with SAR to USD/EUR conversion
Results
Supports all 6 Saudi Riyal denominations (1, 5, 10, 50, 100, 500 SR)
Real-time processing with visual feedback
1,330 reference images for robust matching
Fake note detection using color analysis heuristics
Visualizations
Reference Dataset — Saudi Riyal Denominations
5 SAR
10 SAR
50 SAR
100 SAR
200 SAR
500 SAR