Implementasi Faster Region Convolutional Neural Network untuk Verifikasi Tanda Tangan dan Tingkat Kemiripan Berbasis Mobile

Abstract

Signatures have become one of the important authentication methods in various security and identification applications. However, challenges arise in verifying signatures quickly and accurately, especially in mobile environments. In an effort to overcome these obstacles, this research proposes the implementation of Faster Region Convolutional Neural Network (Faster R-CNN) for signature verification and similarity assessment, especially in mobile platforms. The proposed method combines Deep Learning technology with the high-level object detection capability of Faster R-CNN to recognize signatures in images. Furthermore, it leverages image processing and deep learning techniques to extract features and analyze the similarity between the tested signature and reference data. The implementation is designed to run efficiently on mobile devices, optimizing resource usage while maintaining a high level of accuracy. Performance evaluation was conducted using an extensive signature dataset, including a variety of writing styles and different lighting conditions. Before the verification process, the signature image data will first go through a pre-processing process which includes labeling, resizing, grayscaling, thresholding and continued with feature extraction using canny edge detection. Based on the test results, the system is able to verify signatures with an accuracy rate of 90%