Randomized trial demonstrates accurate blood group classification in real-time using fingerprint images, highlighting innovative deep learning methods.
Key Points
This research aims to develop a deep learning model for automatic blood group classification based on fingerprint images.
Developed a custom Convolutional Neural Network (CNN) to classify 8 blood groups from fingerprint images.
Trained on a balanced dataset of approximately 6,000 images with an oversampling strategy.
Implemented Dropout and MaxPooling layers, optimized training using Adam optimizer with callbacks for efficiency.
Achieved 93% validation accuracy with precision up to 0.98 and F1-score of 0.96.
Classified A+, A-, B+, B-, AB+, AB-, O+, and O- blood groups effectively.
Enabled real-time blood group inference via a Flask REST API for user uploads.