Inspiration
Data mining is an important subject in today's world. however labeled data is a big challenge and sometimes available in a small amount only. Our solution augments existing technology with a machine learning model capable of understanding the items in media frames and labelling them with their corresponding semantics. We believe that applying our solution to CCTV cameras in stores, we can help owners obtain labeled data which can be a useful resource for data miners capable of making crucial suggestions for the owners, like how to design their store in order to enhance their customers' experience overall.
What it does
Using TensorFlow as a machine learning library, our team was able to detect, for this demo, up to 50 classes of elements, in given frames/videos/pictures, including: Person, Cup, Bottle, Table, Chair, etc...
How we built it
As mentioned above, the machine learning library used is TensorFlow. The library uses python as a programming language. Furthermore, we used C# for the Genetec API integration.
Challenges we ran into
Building the application in less than 24 hours was the biggest challenge. Technically, integrating API and mixing programming language introduced minor complexities which we eventually managed to solve.
Accomplishments that we're proud of
The project solves a real life problem and we are happy that it compiles and can be fully demonstrated.
What we learned
Technically, we learned how to integrate Genetec API into a project, which saves us now important steps during future events. Socially, we practiced team work and enhanced our communication skills in a team.
What's next for IntelliCam
Machine Learning is the topic of today. Since our solution is heavy on that latter, new discoveries can easily be adapted into our project. Furthremore, currently the project does not take live footage from CCTV, however this is supposed to work with minor changes if input was provided.
Log in or sign up for Devpost to join the conversation.