GAMIFIED LEARNING PLATFORM FOR INCREASING RECYCLING AWARENESS BASED ON MACHINE LEARNING TECHNIQUES
University POLITEHNICA of Bucharest (ROMANIA)
About this paper:
Conference name: 15th International Technology, Education and Development Conference
Dates: 8-9 March, 2021
Location: Online Conference
Abstract:
Tons of waste are generated in Europe each year and correlated with a growing number of bans and restrictions, coupled with stricter international rules on the export of certain waste to other countries, pollution and garbage remains a very important topic nowadays. There is a lot of concern about the impact of this waste on the environment and our health. Increased resource extraction, production and consumption generates waste that contributes to pollution of air, water, and soil, as well as general climate change and loss of biodiversity. Recycling helps reduce pollution caused by waste and everyone can help make our world a better place if we reuse what we can and recycle what we cannot.
The current article proposes a web platform to educate children how to recycle, by offering attractive multimedia content and interactive rewarding games. The platform is innovative due to its complex technological stack, but most important it uses Tensorflow for implementing machine learning.
Users can bring various objects in front of the computer webcam and the application detects them automatically, using the YOLO (You Only Look Once) algorithm. Then, the user has to drag-and-drop the virtual copy of the object into the correct virtual bin, gaining points for each correct virtual recycling action. Thus, this interactive game was considered attractive by children and not only.
YOLO is an intelligent convolutional neural network (CNN) for performing object detection in real time. The algorithm "only looks once" at the image in the sense that it only requires one pass of forward propagation through the neural network to make predictions. After non-maximal deletion (which ensures that the object detection algorithm only detects each object once), it then generates recognized objects with bounding boxes. Various objects have been used to validate the accuracy of the detection algorithm in the platform, with great confidence, such as: fruits, organic (banana, apple, orange), plastics and metals (laptop, portable, remote control), paper (book), glass (cup, wine glass).
The current study proves that real-time machine learning and image classification in a browser is possible and these modern technologies can be used to educate children and people in general to be aware of the impact of recycling to our wellbeing though an interactive gamified web platform. Keywords:
Educational game, machine learning, YOLO, recycling.