Authors: Yashas D R, Vinutha H N, Merlin B, Soundarya R, Chethan V
Abstract: Identification of an existent's blood group is pivotal in exigency situations, for identity authentication, and in population analysis. It would else involve drawing a blood sample and assaying it in a laboratory, which is painful, tedious and requires trained labor force and installations. Herein, we suggest a way to prognosticate blood groups without blood through the use of point images and a Convolutional Neural Network (CNN). Since fingerprints are distinct in each existent, we suppose they could have patterns associated with natural characteristics similar as blood type. We gathered point images with eight colorful blood groups marked and used them to train a CNN model to classify them. We estimated the performance of the trained model using criteria similar as delicacy, perfection, recall, and F1- score upon testing. Our findings were encouraging, indicating that fingerprints may be potentially employed to cast blood groups using deep literacy. In the future, we will expand our dataset with fresh samples, try out bettered CNN models, and work on securing individualities' data. This system has the implicit to offer an invasive-free, hastily, and easier system for blood group vaticination, particularly in locales with no lab setup.
