Eloc-Web: Uncertainty Visualization and Real-Time Detection of Wild Elephant Locations

The substantial decline in elephant population, primarily caused by human-elephant conflict, necessitates the proactive engagement of conservationists to devise and implement effective monitoring strategies. Monitoring elephants is essential for gaining insights into their movements and ensuring the preservation of habitat corridors. Conservationists have increasingly shifted towards adopting passive acoustic monitoring as an affordable and non-invasive method for determining the spatial distribution of wild elephants through acoustic localization. The main challenge with remote sensing techniques like passive acoustic monitoring is the time-consuming data analysis, which hinders real-time tracking of elephant whereabouts. To address this issue, the study presents Eloc-Web, a web application that visualizes real-time elephant locations using elephant vocalizations recorded by acoustic sensors in the wild, which utilize machine learning to classify the captured audio. Eloc-Web has taken into account the uncertainties associated with classifying captured audio using machine learning models. This ensures that the potential uncertainty of whether the captured sound truly belongs to an elephant is appropriately considered during the visualization process. By following the user-centered design process, the study integrates expert knowledge from elephant ecologists to inform the design and functionality of the application, ensuring its relevance and usability. Eloc-Web, assessed through the System Usability Scale, ranked it in the top 10% of scores, demonstrating above-average user experience and promising potential in assisting elephant ecologists in studying and conserving elephant populations with real-time data visualization.