We are lazy. And not rarely we like to stay in bed on a rainy Sunday. In Finland Sundays are dark. So we need to turn on the lights while we're still in bed, so we can read a nice book.

What it does

Dynalight (TM) is a machine-learning powered real time data fusion all-around service platform for the smart home and company. A heat camera can detect and count persons so you can see the busy spots and easily automize the lighting in your building. Also gestures can be detected and used to control the lights, for example from your bed. With our integrated VR Lightguide (TM) you can guide your guests even when you're not at home or in your workplace.
And that's not all: Dynalight Biofilter (TM) can automatically remove persons and other moving objects from a series of pictures.

How we built it

Here at Dynalight we only work with "brightest" ideas, so our team of experts hacked together this suite solutions in just 2 days, and just for you. So we used an array of tools, including unity, python, processing, a thermal and an hd camera on a raspberry pi and of course smart lights.

Challenges we ran into

We faced many different challenges and learned a ton of stuff during these days: From trying to compile libraries that did not want to, learning to use OpenCV and the interface with the thermal camera, learning to feed images in neural networks, trying to access the smart lights via their REST API and using processing to detect persons in images.

Accomplishments that we're proud of

We actually built a somewhat working prototype that counts persons in a live video feed from the heat camera. While working on many different approaches we also produced a lot of different ideas to interact with the thermal camera.

What we learned

A hackathon is a great place to get to know people, built stuff together and learn from each other. It would be probably great to come with a plan and idea but it's also not necessary since when you start building and trying things out more and more ideas start coming.

What's next for Dynalight

We will extend our ideas and build on our prototypes. A first step would be to have one or more minimal prototypes that we could test with our users. Then, once the gesture detection works, we will stay in bed, switch the lights off and on, remove all persons from our pictures and count the people in our homes and get rich.

Share this project: