Modern society is inundated with advertisements. We see them on billboards. We hear them on the radio. We read them on the newspapers and magazines. And we watch them on the TV shows.
And, these advertisements are more than just annoying. They encourage rampant consumerism, excess pharmaceutical usage, and even help cause childhood obesity by encouraging children to consume sugar filled products.
What it does
Our team decided to take a step towards solving this problem by creating software which detects and then mutes advertisements from live TV video streams.
How I built it
The main mechanism through which our software filters out advertisements is through logo detection. We do some simple preprocessing on the image and try to match it against templates of the channel logos. This is combined with some over-time smoothing to prevent rapid switches to and from muting.
What's next for No More TV Ads!
The main next step would be to design a TV tuner with this software built in. This would enable the easy use of this software and hopefully make an impact by reducing the negative impact advertising has on our society.
An additional further step would be to improve our method for advertisement detection. In particular, we would look into creating a labeled training set and trying to build simple ML models both for the advertising vs show problem and for filtering down the number of possible transition points.