Fork me on GitHub

Mall Camera

Courtesy of the non-commercial CUHK Mall Dataset. Technical details can be found here.

Any input

Count can digest any video feed, web cam, image, image sequence, or sensor data. Real-time or in batches.

Secondary Tooling

We have developed a pipeline for image analysis tooling to extract accurate insights.

Revealing Insights

Count extracts concise, easy to read charts from any organic data source.

Constant Learning

Our feedback mechanic updates our model on a regular basis. The best model is always the next one.

How Crowd Counting at
Shake Shack Works

A diversity of conditions creates a challenging environment. Very tight crowds, blaring sunshine, hard shadows, night, and weather conditions need to be factored into any model. We were able to build our model using over 10,000 ground truth images.

Live Input

Shake Shack provides a live web camera we can pull from every 5 minutes.

Mapping Layer

We create a density map to get an accurate crowd count before we deduce the number of people in line.

Accurate Data

We constantly compare ground truth information to our latest prediction to build our model.

Count builds image analysis for businesses.

Get in touch

Get in touch: @dimroc.

Count is a partnership between Evolution of Books LLC and Thought Merchants.