Overview / Usage

​

People with visual impairments face significant challenges in their daily lives, struggling with tasks like navigation, object recognition, and accessing information. These challenges often lead to dependence on others, reduced autonomy, and limited access to the wealth of information available in the modern world. BlindMate addresses these issues by harnessing the power of advanced AI and computer vision technologies to provide a comprehensive solution that enhances the lives of visually impaired individuals. BlindMate employs AI-powered object recognition to accurately identify everyday objects, enabling users to easily differentiate between items and interact with their surroundings more effectively. The app's advanced feature of recognizing currency notes assists visually impaired users in managing their finances independently, ensuring they can confidently handle money-related transactions.

Methodology / Approach

​

Our methodology for addressing the challenges faced by visually impaired individuals through the creation of BlindMate is centered around a strategic integration of advanced technologies, frameworks, and techniques. By effectively combining these elements, we are able to provide practical solutions that enhance the lives of visually impaired users.

Firstly, we selected the Android platform as our foundation, leveraging its widespread accessibility and user familiarity. We then incorporated Intel's OneAPI library, a powerful framework known for optimizing and accelerating workloads across diverse Intel architectures. This choice enables us to achieve real-time processing and performance optimization, crucial for delivering instantaneous assistance.

Our development heavily relies on deep neural networks (DNNs) trained with datasets, allowing us to implement complex computer vision techniques such as object detection. These techniques empower BlindMate to process live camera feeds, enabling real-time scene analysis, accurate object recognition, and even currency note identification. Through a rigorous training process, we fine-tuned our DNN model to ensure reliable and precise predictions.This comprehensive approach enables us to provide an innovative solution that empowers the visually impaired community, fostering independence and inclusivity in their daily lives.

Technologies Used

Technologies and Libraries:

• Android Platform - The operating system used for developing the mobile app.
• Intel OneAPI Library - A versatile framework for optimizing and accelerating workloads across Intel architectures.
• OneDNN - Advanced machine learning models used for object recognition, scene analysis, and currency note identification.
• Object Detection Algorithms - Techniques for identifying and locating objects within images or video streams.

Software and Tools:

• Android Studio - Integrated development environment (IDE) for Android app development.
• TensorFlowLite and PyTorch - Popular frameworks for building and training deep learning models.
• Git - Version control system for collaborative development.

Hardware:

• Android Mobile Devices - Smartphones running the Android operating system for app testing and deployment.
• Cameras - Built-in or external cameras used to capture visual information for processing.

Documents and Presentations

Repository

https://github.com/JasleenKaur2711/BlindMate

Collaborators