Deep learning workflow to support in-flight processing of digital aerial imagery for wildlife population surveys.

Deep learning shows promise for automating detection and classification of wildlife from digital aerial imagery to support cost-efficient remote sensing solutions for wildlife population monitoring. To support in-flight orthorectification and machine learning processing to detect and classify wildlife from imagery in near real-time, we evaluated deep learning methods that address hardware limitations and the need for processing efficiencies to support the envisioned in-flight workflow. We developed an annotated dataset for a suite of marine birds from high-resolution digital aerial imagery collected over open water environments to train the models. The proposed 3-stage workflow for automated, in-flight data processing includes: 1) image filtering based on the probability of any bird occurrence, 2) bird instance detection, and 3) bird instance classification. For image filtering, we compared the performance of a binary classifier with Mask Region-based Convolutional Neural Network (Mask R-CNN) as a means of sub-setting large volumes of imagery based on the probability of at least one bird occurrence in an image. On both the validation and test datasets, the binary classifier achieved higher performance than Mask R-CNN for predicting bird occurrence at the image-level. We recommend the binary classifier over Mask R-CNN for workflow first-stage filtering. For bird instance detection, we leveraged Mask R-CNN as our detection framework and proposed an iterative refinement method to bootstrap our predicted detections from loose ground-truth annotations. We also discuss future work to address the taxonomic classification phase of the envisioned workflow.

Patterns of mercury and selenium exposure in minnesota common loons.

Common loons (Gavia immer) are at risk of elevated dietary mercury (Hg) exposure in portions of their breeding range. To assess the level of risk among loons in Minnesota (USA), we investigated loon blood Hg concentrations in breeding lakes across Minnesota. Loon blood Hg concentrations were regressed on predicted Hg concentrations in standardized 12-cm whole-organism yellow perch (Perca flavescens), based on fish Hg records from Minnesota lakes, using the US Geological Survey National Descriptive Model for Mercury in Fish. A linear model, incorporating common loon sex, age, body mass, and log-transformed standardized perch Hg concentration representative of each study lake, was associated with 83% of the variability in observed common loon blood Hg concentrations. Loon blood Hg concentration was positively related to standardized perch Hg concentrations; juvenile loons had lower blood Hg concentrations than adult females, and blood Hg concentrations of juveniles increased with body mass. Blood Hg concentrations of all adult common loons and associated standardized prey Hg for all loon capture lakes included in the study were well below proposed thresholds for adverse effects on loon behavior, physiology, survival, and reproductive success. The fish Hg modeling approach provided insights into spatial patterns of dietary Hg exposure risk to common loons across Minnesota. We also determined that loon blood selenium (Se) concentrations were positively correlated with Hg concentration. Average common loon blood Se concentrations exceeded the published provisional threshold. Environ Toxicol Chem 2019;38:524-532. Published 2018 Wiley Periodicals Inc. on behalf of SETAC. This article is a US government work and, as such, is in the public domain in the United States of America.