Awards & Nominations
Praxidike has received the following awards and nominations. Way to go!
Global Nominee
Crowd density estimator
High-Level Project Summary
Our project is an app which is meant to give people an idea of how crowded an area they might want to visit in their city is. This is achieved through a map which displays common places that people visit as well as the density of those areas using our neural network model for classifying three different density levels. In this way, people can easily check how crowded the area they want to visit is and avoid any high-density spots, so as to limit the spread of COVID-19. Along with other quality-of-life features, the importance of this project comes down to protecting the citizens' health as well as making progress in the battle against COVID-19.
Link to Project "Demo"
Link to Final Project
Detailed Project Description
Our project uses convolutional neural networks to classify three different levels of density. This is combined with our website and map which make up the graphical-user interface of the application.
The data-set we used was completely compiled by us from copy-right free images from the internet.
For the model, we first convert the images from the data-set into a smaller size format and split them into training and validation arrays. Next up, we structure the neural network using convolutional, pooling and dense layers from Keras, as well as data-augmentation and Dropout layers to decrease the overfitting of our model. Then, we fit our model with the training data and test it on the validation data, which provides better accuracy on images not used during training. The model classifies only high and low levels of density, but we create a third middle level of density artificially using the values from the other two levels.
The benefits of our project include the easy-to-use graphical interface, so that anyone who wants to check the density of a given area can do so easily and quickly.
We hope to achieve a lower-rate of COVID-19 infections by helping people avoid densely populated areas.
During development we used Python with Tensorflow, Keras and Scikit-learn for the model, HTML5 CSS3 JS6 (with Leaflet.js, Stadia and Freshdesk) for the website.
Space Agency Data
We have not used any space agency data in our project.
Hackathon Journey
We would describe our experience as very educational, albeit rough at times. We learned how to work in a team and build something from scratch using our own knowledge and skills. We were inspired to choose this challenge with the thought of helping our society overcome the pandemic, so we can finally return to the good old times. We have consulted with our teachers, as well as researching all the necessary technology we needed to build the application. We resolved setbacks and challenges with hard work and dedication. Our team would like to thank all of our teachers who have supported us throughout the years and made it possible for us to take part in this Hackathon.
References
Tags
#covid #ai #tensorflow #javascript #crowd-estimation
Global Judging
This project has been submitted for consideration during the Judging process.