The cost of living in Singapore has increasingly become a central concern for individuals, businesses, and government agencies. Despite the abundance of economic data available, much of it remains fragmented, difficult to interpret, and lacking the tools needed to help the public understand how broader economic trends affect everyday life. Currently, there is no unified platform that consolidates cost-of-living data, income information, and predictive insights tailored specifically to the Singaporean context.
CostInsightSG was developed to address this gap. It is a cloud-based web application that integrates official Singapore government datasets and transforms them into accessible, interactive visualizations. By combining Consumer Price Index data, income statistics, and historical trends, the platform enables users to better understand how their personal finances relate to the broader economic landscape. Beyond displaying current trends, it also offers forecasting features that allow users to anticipate future changes in living costs.
In recent years, the rapid advancement of technology has significantly elevated customer expectations within the travel and airline industry, particularly for efficient and convenient service. In response to this demand, this project presents the development of an intelligent, user-friendly chatbot designed to enhance and streamline the customer experience. This chatbot is crafted to simplify and accelerate essential customer interactions, and provide users with a comprehensive tool to manage their travel arrangements with ease and efficiency.
We have designed and implemented this chatbot as a full-stack application, incorporating both a front-end webpage and a back-end database in order to simulate real user experience. By interacting with the chatbot, users can manage their bookings and address common needs directly in a single, accessible interface. Through this project, we aim to enhance the airline’s customer service by making travel management simpler, faster, and more user-centric.
This project introduces a Vision Transformer (ViT)-based method for recognizing twelve specific characters from the game Genshin Impact in cosplay images. We collected a dataset by web-scraping 150 images for each character, covering both in-game character models and cosplayer representations. To better isolate the characters in each image, we used YOLOv8 human detection to accurately identify and crop character regions to minimize background interference. After preprocessing, we apply data augmentation techniques to diversify the dataset, enhancing the model’s resilience to different cosplay styles and environments. The augmented dataset is then used to train and fine-tune a pre-trained ViT model. Our experiments show that the ViT-based model can reliably identify Genshin Impact characters in cosplay images, underscoring the effectiveness of combining advanced deep learning models with rigorous data preparation for character recognition tasks. The model achieved an accuracy of 88%, which is notable given the relatively small dataset it was trained on and is suitable for inference on both images and video, demonstrating high practical usability.