Advanced logo recognition

Inspiration

Rohde & Schwarz asked us to programm an television logo recognizer to detect the logos in television frames.

What it does

We are filtering the high frequencys out of the picture and search for equality with our reference pictures of the logos-

How we built it

Challenges we ran into

Some pictures are so noisy that our solution got in trouble wo distinguish if there is a logo or not. For a human it wasn't even possible to see the logo in some cases.

Accomplishments that we're proud of

We are able to detect the logo in most pictures. Our build from scratch neural network is able to learn. Managing meridian Filtering the pictures.

What we learned

A good picture resolution is needed to recognize logos with high accuracy.

What's next for Advanced logo recognition

We will improve our UI and merge our actual logo recognition with our neural network.

Built With

• c++