Urban Metrics
Our urban neighborhoods need attention. Innocent pedestrians are getting hurt as they cross our roadways due to poor traffic control. We need an Android application that will allow us to capture objects using computer vision to count both pedestrians and traffic objects within urban environments.
Solving urban traffic needs utilizing Android and Computer vision to detect traffic and pedestrian metrics. Example use-cases included four-way streets, two/one-way traffic, and parking garages
Role: Lead architect, technical project manager, and backup engineer.
Impact: The codebase was sold to Google Labs.
|
Collect Sample Videos |
|
Manually Count Cars & Pedestrians in Each Video to compare results and train |
|
Implement OpenCV on Android using the NDK |
|
Compare and test detection methods including blob detection and feature detection |
|
Save detected objects and log events locally |
|
Filter to differentiate pedestrians vs cars and remove false positives |
|
Project Documentation |
|
Accuracy Testing and Reporting |