Yan Tan Tethera

Inspiration

There are many usages for a people counting system . It could be used for counting no. of people in a festival / protest / shopping mall. The collected data is good for analysis.

Thermal camera is a perfect match with people counting system.

However, we have a question during brainstorming. Could it do more than counting people? Could it be more fun?

Finally, we find that it could also be used to count mouse in kitchen in a restaurant.

Of course people will try hard to keep kitchen be clean, but mouse may come from outside. Therefore, People may use a counting system to monitor any small animal entry your kitchen, and take action to solve it.

Yan Tan Tethera is designed for counting different living things for different needs.

What it does

Features

Counting no. of people walk through a place
Counting no. of mouse walk through your kitchen / restaurant

How I built it

We use Flir One Camera to collect the thermal image. Then process the result image by OpenCV. The algorithm for counting people and mouse is slightly different. But generally can be divided into the following steps.

Perform a background removal on the input frame to finding out moving objects
Apply a range filter of temperature if needed
Apply a blob detector
Apply blob tracking algorithm
Count no. of unique blob

Challenges I ran into

Since we cannot demonstrate our project during pitching due to installation problem, we decided a take video and lets Yan Tan Tethera to read from the video file. However, OpenCV on Android do not support video file reading. We need to write our wrapper and it is very time consuming.