Aerial Surveys + AI for Accurate Wildlife CountsWild Me Scout is an open hardware and open source software solution designed to support analysis of large volumes of data that are obtained from aerial surveys of wildlife. By using photos, we can make aerial surveys cheaper, more accurate, and safer.
|
Aerial surveys are a well established tool for wildlife population estimation, but they come with a number of problems. If taken from a plane with rear seat observers, planes have to fly low and slow, and observer fatigue can impact the counts obtained. To improve safety, you can replace rear seat observers with cameras. However, whether by plane or by drone, camera-based surveys produce massive amounts of imagery, which can number into the hundreds of thousands of photos. Most of these images do not contain the relevant species, and the cost of manually annotating and processing for analysis can make this an expensive, time consuming, and ultimately nonviable process.
Scout is a multi-user software application that supports parallel human and machine learning (ML) review of aerial survey imagery. Scout's two-part ML process quickly throws out images that do not contain the relevant species under study (often > 98% of imagery) and then annotates (bounding boxes) and labels the species for animals detected in the remaining images.
With Scout, you can:
Scout is a multi-user software application that supports parallel human and machine learning (ML) review of aerial survey imagery. Scout's two-part ML process quickly throws out images that do not contain the relevant species under study (often > 98% of imagery) and then annotates (bounding boxes) and labels the species for animals detected in the remaining images.
With Scout, you can:
- Ingest high volumes of .ARW and .JPG images collected from survey cameras
- Group images as “Tasks” that can be assigned to other users (e.g., image annotators) or machine learning (ML) models for bounding box creation and species labeling
- Review and “ground truth” annotated images for accuracy
- Draw division lines for overlapping image sequences with annotations
- Export CSV data files for statistical analysis