Semester 2 2024 Web Development
CARE Web Platform: Advancing Wildlife Conservation through AI-Driven Animal Re-Identification
Description
New Zealand is home to some of the world's most unique fauna, but these native species face significant threats from invasive pests. One such predator is the stoat, which preys on the young of native birds. Our client has recognised this issue and tasked us with developing a web platform to assist in monitoring and controlling the presence of stoats on Waiheke Island. The platform leverages machine learning and artificial intelligence to accurately identify stoats and other animals captured in photos taken across the island. By utilising advanced image recognition algorithms, the platform can differentiate between various species, ensuring precise identification and effective monitoring. This innovative approach not only enhances the efficiency of pest control efforts but also contributes to the broader goal of preserving New Zealand's unique biodiversity.
Details
Contributors Kingsley Leung (Team leader, Full-stack Developer) Zhihong Yang (UX/UI Designer, Front-end Developer) Kevin Yao (UX/UI Designer, Front-end Developer) Yichen Liu (Back-end Developer) Franklin Yu (Back-end Developer) Xiaolin Zhao (AI Deployment, Back-end Developer) Collaborators & Acknowledgements We would like to extend our heartfelt thanks to Zhao Di, Yun Sing Koh and Justin Wu for their invaluable expertise and support in integrating our platform with their CLIP-based AI model. Their insights and guidance were crucial in navigating the complexities of this integration, helping us enhance the functionality and performance of our application. Also, special thanks to Anna Trofimova, Asma Shakil, and Matthew Alajas for their ongoing instruction and guidance. Moreover, thanks to the team Poke Rangers from the University of Auckland's COMPSCI 399 Capstone course (2024 S1). This project utilises some of their frontend code. Future Plan One area for improvement is to refine the progress bar for different processing tasks. Our current implementation displays an attractive progress bar that shows the number of processed images, the total number of images, and the percentage completed. However, it would be beneficial to include an estimated time remaining, which would help users understand how long they will need to wait. There are some areas for improvement in the frontend interface. For instance, the user guide is currently located at the bottom of the page, which may not be easily noticeable for all users. It would be more effective to move the link to a more prominent position, such as the top navigation bar. Furthermore, it would be beneficial if the image gallery could display the metadata of the currently previewed image. Typically, images contain useful EXIF data, such as GPS coordinates and the time they were taken. By providing this metadata, users will gain a better understanding of all images and their contexts, which will enhance data analysis.
developed by BinaryBuilders
No comments yet, Start the conversation.