Screenshot of the tool

Inspiration

For years, marketing departments have tried to understand and measure people's reaction to a given product. For this purpose, many ways of measuring popularity and attraction have been developed. Unfortunately, one component has biased them heavily - almost all previously used measurements depend on people evaluating themselves, or someone else's subjective opinion. Simply, human error is involved in the process.

Our solution eliminates human subjectiveness from the equation, allowing for an objective analysis of the impact of the product/speech on the audience. Using Crowdminator one can obtain a direct, precise and unbiased metric of audience emotions in a very short amount of time. This solution opens a whole new world for the opinion mining industry.

Applications and impact

Our product can be applied in any area where an enhanced audience analysis is required, among the applications of the tool we can highlight:

Digital Marketing, Product Development.
Politics, Big Brother, 1984.
Movies, theaters, stand-up comedy, tv-shows, hackathons...

Implementation

The tool is implemented in Python and it makes use of Microsoft's Face API to make the face detection and sentiment analysis of the people in the images. Furthermore, it processes the data and creates an interface to visualize the results.

How does it scale ?

The solution can be applied to any video or set of images. Current Microsoft's Face API limitations constrain its applicability to real time and limit the number of people in the video, but these constraints are imposed and could be removed. Therefore the solution can be applied to virtually any video.

Future Development

In the tool you can specify which feelings are considered of interest (positive) and which ones are not. Future developments lines would be:

Tracking of the emotions of specific members in the crowd and their relationships with others.
Combination with sound information.
Real time application.
Automatic report generation and anomalies detection.
Apply cultural filters (Finnish people would statistically smile less than a Spaniard).