Cat Detective Round 1: Trying OpenCV

I’m both fascinated and creeped out by Facebook, Google, and similar technology companies whose platforms consume and analyze the entire world’s personal data. After seeing the recent Google app that attempts to match people’s faces to famous works of art, I was curious to see how difficult it would be to learn how to use and apply this technology myself.

My starting point was a search for “open source facial recognition” which brought me to a project called OpenCV, which ultimately led me to a wonderful series of tutorials on Adrian Rosebrock’s web site, Pysearch.

OpenCV and Rosebrock’s tutorials use Python, which I don’t have any experience with. Even so, I followed his directions and was able to set an OpenCV environment without much trouble and got my feet wet with Rosebrock’s Hobbits and Histograms” tutorial within a few hours. I’m not interested in an image search engine, so I quickly moved on to the cat detector.

Playing Cat Detective: Round 1

Using Rosebrock’s Detecting cats in images with OpenCV tutorial.I successfully recreated Rosebrock’s results on three images:

Test file #1

Success! Correct number of cats detected.

Test file #1 Test file #1

Test file #2

Success! Correct number of cats detected.

Test file #2 Test file #2

Test file #3

Success! Correct number of cats detected.

Test file #3 Test file #3

This looks like it will be interesting to play with. There are lots of configuration options and different algorithms available for use with OpenCV. In the short-term, I’ll be sticking with Rosebrock’s configuration and see where that takes me.

Next Steps

In Round 2 I will attempt to answer the question: “Can the cat detector find cats in photos my friends and I take?”