OUTSIDE BIN (yesterday)
The AI recognising the coke can.
The AI recognising a cup
the sorting machine
The AI recognising a packet
where'd u get that damn good idea from?
Our inspiration comes from an unlikely source- MCDONALDS. While we were sitting there, discussing ideas for this hackathon, we saw the table across from us, full of trash and waste (most of which could be recycled). IN FACT (look at one of the images) the we ourselves (the participants of this hackathon) didn't recycle our biodegradable plates!
YO, what is the problem ur tryin' to solve?
after some digging around we realized how big of a problem this actually is, especially in Singapore! As it turns out Singapore only recycles <24% of domestic waste this is VERY LOW compared to the US (35% recycle rate) and Germany (87% recycle rate). We also saw that people SELDOM put the correct material in the correct waste bin. Which
Cool, what does it do?
Using CoreML and Computer Vision, connected to a Wemos Mini Wi-Fi connected board, we have created a simple to use sorting system, combined with an app interface that sorts waste into different bins. Through image classification, we have created a way to identify, and count the amount and type of waste consumed by household. By feeding this information into a database, waste management organizations can optimize the path to only visit full trash bins, thereby saving on fuel.
OMG, How in the world did u make it?
For the AI, we used a computer vision model and we edited the detection, and added a web request to easily connect to the sorting system – a simple servo connected to a Wemos Mini board. We built the hardware using just plain old cardboard and glue, making this cheap and easy to use. Using a database hosted on a cloud, the path calculation is a simple app interface that can be accessed by any of the truck drivers available.
But it wasn't all smooth sailing, was it?
Of course not! There were many ups-and downs and challenges that we faced. The most important one being that our team-mate (Sarvasv) who was responsible for building the AI, got sick on Saturday, coming as a major set-back. This is the primary reason why our product is not as polished as we'd hoped, but we believe that the idea is more important than the prototype (for the hackathon)
Hmm, what are u most proud of in this?
We are very proud of our AI that allows for the trash detection to be so simple yet effective – bringing complex sorting machines into everybody’s homes. But most of all, we are proud of our idea.
It must have been a jolly good learning experience right?
So watch'ya goona do next with this?
we actually see this as a very real product, as it was designed with a very real problem in mind. Each aspect of the design has been thought out such that it can be implemented easily and efficiently into the industry.