Our proprietary algorithm utilizes our training data to recognize urban planning assets. We are building out our pipeline to expand the accuracy and capability of our model; including sizing each asset accurately. The image above shows the STRATA's algorithm recognize travel lanes, parking lanes, and sidewalks.
Our Machine Learning Model is using a proprietary methodology to tag existing data and identifying design assets. STRATA is on its way to capture its own data. At the moment, we our expanding its recognition capability for sharrows, bike lanes, and detailed lane markings.
Project Third Eye
The image analytics effort under STRATA uses various methodologies in machine learning and neural nets; toolkits primarily applied in the autonomous vehicle space. We are currently building our data capture pipeline as well as working on strengthening the algorithm to recognize the urban design / infrastructure assets. The image below is the work of machine learning methodologies (AI to the general public) that can differentiate various infrastructure assets (ie: travel lanes, bike lanes, parking lanes, sidewalks, buffers, etc).
The images on this page are various corridors in NYC that STRATA trained its machines learning models on. We are excited to put our model through its paces with various degrees of infrastructure complexity.
The image above is E. 222nd Street in the borough of Bronx in New York City. It is a residential corridor with various treatments across its length. The highlighted areas are the work of an ML model learning to recognize assets in street design. This is proprietary to STRATA.