Every year, the production of waste increases exponentially. Moreover, it is estimated that humankind currently produces two billion tonnes of waste per year between 7.6 billion people. The people that are affected most about this trend will be the generations that come after us. (https://sensoneo.com/sensoneo-global-waste-index-2019/)
As of right now, recycling is still the best method for managing waste and so we have developed an educational app that will encourage the younger generation to recycle. Our goal is to promote environmental awareness to the generations that come after us. Using AI image classification, you are able to take a photo with your mobile device and it can determine whether something is recyclable or not. To target the younger generation, our aim is to partner with big organizations to sponsor the act of recycling. Similar to how the LCBO reimburses for alcohol cans, we can partner with government organizations or other companies who want to carry out environmental initiatives. So people who take pictures of recyclable items will be rewarded currency (maybe even companies like Cineplex for scene points?)
Technologies used are Kotlin, Android Studio, Python, Keras, Tensorflow, Microsoft Azure and Firebase
The learning curve for Kotlin and Android Studio was larger than we expected but we managed to push through. Also, integrating the AI posed to be a very difficult challenge.
The major accomplishments that I am proud of would be the process of data scrapping over 6000 images off Google's search engine to feed into our Convolutional Neural Network. As well, our team is proud of training this neural network to an accuracy of 92.5% via using a method called retraining.
Within cuHacks, I learned how to create a dataset from scratch, train an neural network and work within Kotlin
The next steps for ecoSNAP are to configure the HTTP requests with a live server, loading a compressed ai directly onto the application to reduce latency, and to feed more data into our algorithm to distinguish more objects and how to appropriately dispose of them. As well, we hope to create small little mini games and quizzes for our users to learn and gain additional points.
