Abstract

Object detection plays a crucial role in various real-world applications, including assistive technology for the visually impaired. This research presents a deep learning-based object detection and recognition framework utilizing SSD300 (Single Shot MultiBox Detector) and InceptionV3 to enhance object classification, specifically for currency note recognition. The integration of these models allows the detection of 22 object classes, including currency notes, achieving an accuracy of over 98%.The proposed framework addresses the limitations of traditional object detection models by improving feature extraction and classification accuracy. The hybrid SSD-Inception model leverages SSD300’s speed and InceptionV3’s deep feature learning capabilities, ensuring robust detection performance. Extensive experiments were conducted using a dataset of Indian old currency notes, demonstrating significant improvements in classification accuracy. The system is designed to assist visually impaired individuals by accurately recognizing and categorizing objects in their environment. The findings indicate that deep learning models, when integrated effectively, can provide highly reliable real-time object detection solutions. Future advancements may include the incorporation of new currency datasets, real-time implementation, and deployment on edge devices for enhanced accessibility.