A computer vision for animal ecology

Crop diseases are a major threat to food security, but their rapid identification remains difficult in many parts of the world due to the lack of the necessary infrastructure. The combination of increasing global smartphone penetration and recent advances in computer vision made possible by deep learning has paved the way for smartphone-assisted disease diagnosis. Using a public dataset of 54,306 images of diseased and healthy plant leaves collected under controlled conditions, we train a deep convolutional neural network to identify 14 crop species and 26 diseases (or absence thereof). The trained model achieves an accuracy of 99.35% on a held-out test set, demonstrating the feasibility of this approach. Overall, the approach of training deep learning models on increasingly large and publicly available image datasets presents a clear path toward smartphone-assisted crop disease diagnosis on a massive global scale.

The behavior of individuals determines the strength and outcome of ecological interactions, which drive population, community, and ecosystem organization. Bio-logging, such as telemetry and animal-borne imaging, provides essential individual viewpoints, tracks, and life histories, but requires capture of individuals and is often impractical to scale. Recent developments in automated image-based tracking offers opportunities to remotely quantify and understand individual behavior at scales and resolutions not previously possible, providing an essential supplement to other tracking methodologies in ecology. Automated image-based tracking should continue to advance the field of ecology by enabling better understanding of the linkages between individual and higher-level ecological processes, via high-throughput quantitative analysis of complex ecological patterns and processes across scales, including analysis of environmental drivers. Copyright © 2014 Elsevier Ltd. All rights reserved.

Aerial surveys of marine mammals are routinely conducted to assess and monitor species’ habitat use and population status. In Australia, dugongs (Dugong dugon) are regularly surveyed and long-term datasets have formed the basis for defining habitat of high conservation value and risk assessments of human impacts. Unmanned aerial vehicles (UAVs) may facilitate more accurate, human-risk free, and cheaper aerial surveys. We undertook the first Australian UAV survey trial in Shark Bay, western Australia. We conducted seven flights of the ScanEagle UAV, mounted with a digital SLR camera payload. During each flight, ten transects covering a 1.3 km2 area frequently used by dugongs, were flown at 500, 750 and 1000 ft. Image (photograph) capture was controlled via the Ground Control Station and the capture rate was scheduled to achieve a prescribed 10% overlap between images along transect lines. Images were manually reviewed post hoc for animals and scored according to sun glitter, Beaufort Sea state and turbidity. We captured 6243 images, 627 containing dugongs. We also identified whales, dolphins, turtles and a range of other fauna. Of all possible dugong sightings, 95% (CI = 90%, 98%) were subjectively classed as ‘certain’ (unmistakably dugongs). Neither our dugong sighting rate, nor our ability to identify dugongs with certainty, were affected by UAV altitude. Turbidity was the only environmental variable significantly affecting the dugong sighting rate. Our results suggest that UAV systems may not be limited by sea state conditions in the same manner as sightings from manned surveys. The overlap between images proved valuable for detecting animals that were masked by sun glitter in the corners of images, and identifying animals initially captured at awkward body angles. This initial trial of a basic camera system has successfully demonstrated that the ScanEagle UAV has great potential as a tool for marine mammal aerial surveys.