Visual impairment is a widespread problem that affects millions of people worldwide. According to the WHO, over 2.2 billion individuals have some level of vision impairment, with at least 1 billion being visually impaired. To enhance mobility and independence, visually impaired individuals rely on tools like white canes, guide dogs, and assistive technologies. However, there is a need to transform the visual world into a sound-based one, providing information about objects and their spatial locations. Deep learning algorithms have made significant advancements in the field of computer vision, specifically in object detection. Convolutional neural networks (CNNs). There are two main types of approaches for object detection in deep learning: two-stage and single-stage algorithms. Two-stage algorithms, like R-CNN, Fast R-CNN, and Faster R-CNN, use a regional proposal network to identify regions of interest. In the second stage, these regions are classified and their locations are predicted. This two-stage approach has its own great potential in enhancing the accuracy of object detection. That’s where our proposed system comes in. We aim to develop a compact, cost-effective solution using computer vision to detect and recognize objects for visually impaired individuals. By utilizing pre-trained datasets, we can convert object information into text and then convert that text into speech. The way you have incorporated the SSD mechanism and the Inception v3 model into your framework is truly impressive. Standalone SSD detector, and testing accuracies of 92.7% for person recognition and 90.4% for currency recognition using Inception v3.