Computer Vision for impaired individuals

• Category: Deep Learning & Computer Vision
• Context: MSc team project, IE University
• Project date: March 2023

Description:

VisionAssist is an Android application designed to aid visually impaired individuals by detecting objects in their environment using YOLOv5. The app uses a custom-trained YOLOv5 model, which identifies 14 common street objects, helping users navigate safely. The model, trained on a dataset of 730 annotated images, is optimized with ONNX, OPENVINO, and TFLITE for efficient deployment on Android devices. Future enhancements will include text-to-speech functionality for real-time auditory feedback.