Trailcam Data

Online Platform:

Through conversations with wildlife researchers and managers, the problem of data collection became evident. More specifically, the popularity of camera traps has been leading to an increase in footage developed which has increased work required while staffing hasn’t increased. The leading cause of the increased workload was the exponentially increasing rate of false positives, reaching upwards of 30%. Leading a crossfunctional team, we partnered with organization to collect needed to develop the CNN model (reaching a 95% accuracy), and developed a file sorting tool to bucket footage of animals and false positives.

Sorting Tool:

With success from the false positive product, there was growing demand for an online tool that could provide smaller organizations with a system where users could upload wildlife footage, sort, organize, and analyze. Through discussion around current problem areas, the design of the online platfiorm would service a different set of needs than the false positive removal system. The platform would have to provide more features to track wildlife activity and determine population and migration trends. Using the machine learning model from the false positive removal system integrated through a JSON framework an online platform was developed. Data was fetched using python to aggregate data and google visualizations to tabulate the data.