Problem

A group of developers wanted to create a solution that would help visually impaired people navigate their surroundings with ease. They realized that there was a need for a system that could recognize locations based on images, and provide audio cues to the user.

However, they faced several challenges in creating this system. Firstly, they needed to create a robust training dataset that would enable the system to recognize different locations accurately. Secondly, they needed to develop sophisticated algorithms that could classify images and provide accurate location information.

Watch our demo video here!

Solution

The team decided to use Python, Computer Vision, and Machine Learning technologies to create their solution. They started by collecting a large dataset of 2700 photos of different locations using Google Street View. They then used K-means clustering techniques to group similar images together and remove duplicates.

Next, the team used Decision Trees, Support Vector Machines (SVMs), and K-nearest neighbor (KNN) algorithms to classify the images based on their features. This resulted in superior place recognition capabilities with a 94% accuracy rate.

Impact

Improved navigation capabilities for visually impaired individuals
Increased independence for visually impaired individuals
X% increase in accuracy rate compared to existing solutions

"This solution has had an incredible impact on the lives of visually impaired individuals who can now navigate their surroundings with ease."

Team

Team Member #1 - Software Engineer
- Oversaw the development of the algorithms and classifiers
- Implemented K-means clustering techniques to group similar images together
Team Member #2 - Data Scientist
- Created the training dataset of 2700 photos using Google Street View
- Used Decision Trees, SVMs, and KNN algorithms to classify the images based on their features

Developers Create 94% Accurate Navigation for Visually Impaired