Environmental data collection from sensor infrastructure for smart nation project

Abstract

Data annotation is the process of labelling data to provide the desired outcome that the model should predict. This process usually requires an exorbitant amount of cost and effort if the annotation is done by human annotators. In this project, an automated data labelling model will be implemented to predict the event onset and offset. The network architecture to be implemented consists of pre-trained VGG16 together with bi-directional GRU layer. The model performance will be using event-based macro F1 score, intersection over union, precision and recall for evaluation