The fake news detector detects whether a social media post contains fake news. One of the main features was the stance detection algorithm that was done using various NLP tools and deep learning models such as BERT and roBERTa. The model was integrated into a web application built using React to provide a platform for users to easily use the fake news classifier. The classifier achieved a high f1 accuracy of 80%.

We helped our end users to shorten the time taken to review assessment scores of their candidates by introducing interactive dashboards done using Tableau to replace the standard hardcopy reports for each candidate. Users can measure the performance of a candidate through visual clues and can easily compare an individual’s performance to the cohort.

We also used machine learning models such as logistic regression, random forest, naive bayes and neural network to do a predictive modelling for each candidate which we believe will help to speed up the process of shortlisting candidates based on their performances.

We built a python GUI to help the end users to easily train and use the models.

To aid the safe distance ambassadors in looking out for those who are not wearing a mask or wearing the mask incorrectly, thereby increasing productivity and efficiency.

This real-time face mask detector is built using Keras, Tensorflow, MobileNet and OpenCV.