P02-week2-Data Seeking and Analysis of Existing Data
In the heart of conservation efforts and ecological research, camera trap data has emerged as a powerful tool for monitoring wildlife in its natural habitat. These devices capture images or videos when triggered by an animal’s movement, providing researchers with a wealth of information about biodiversity and ecosystem dynamics. However, before delving into complex analyses and modeling, it is essential to conduct a thorough Exploratory Data Analysis (EDA) to understand the nuances of the captured data. In this blog post, we will embark on a journey through the wilderness of camera trap data, exploring the insights of data available on the net.
Camera trap datasets often consist of thousands or even millions of images, each containing a timestamp, location, and potentially additional metadata. Before jumping into analysis, it’s crucial to grasp the basics of the data structure and format. So we have started by trying to understand what data offers.
Problems
The first problem we came across is that each project has its own labeling standards so if we are thinking of merging multiple datasets from different projects into one, we have to consider this issue.
The second issue is that we want our model to be as inclusive as it can be for endangered species. So filtering out data for these species will be time-consuming. Currently, we are working on datasets from https://lila.science/datasets. This website offers more than 30 projects gigabytes and even terabytes of images. So we think this website might come in handy but we need to work on acquiring the most satisfying dataset.
Conclusion
To acquire a dataset that is suitable for our purpose, we need to spend most of our time gathering & merging applicable camera trap images without getting distracted by cute animals caught off-guard.
