Recent technological advancements have seen a rapid growth in the use of imagery data to estimate the abundance and spatial distribution of animal populations. However, the value of imagery data may not be fully exploited under traditional analytical frameworks. We developed a method that leverages aerial imagery data for population modeling through entity resolution, a technique that stochastically links the same individual across multiple images. Resolving duplicate individuals in overlapping images that are distorted requires realigning observed point patterns optimally; however, popular machine learning algorithms for image stitching do not often account for alignment uncertainty. Moreover, duplicated individuals can provide insight about detection probability when overlaps are viewed as replicate surveys. Our model resolves individual identities by linking observed locations to latent activity centers and estimates total population as informed by the linkage structure. We developed a hierarchical framework to achieve entity resolution and abundance estimation cohesively, thereby avoiding single-direction error propagation that is common in two-stage models. We illustrate our method through simulation and a case study using aerial images of sea otters in Glacier Bay, Alaska. Supplementary materials accompanying this paper appear on-line

最近的技术进步已经看到使用图像数据来估计动物种群的丰度和空间分布的快速增长。然而，图像数据的价值可能无法在传统的分析框架下得到充分利用。我们开发了一种方法，该方法利用航空影像数据通过实体分辨率进行人口建模，这是一种随机链接多个图像中的同一个人的技术。解决重叠图像中扭曲的重复个体需要以最佳方式重新对齐观察到的点模式；然而，用于图像拼接的流行机器学习算法通常不能解释对齐的不确定性。此外，当重叠被视为重复调查时，重复的个体可以提供有关检测概率的见解。我们的模型通过将观察到的位置与潜在活动中心联系起来来解决个体身份，并根据联系结构估计总人口。我们开发了一个层次框架来实现实体分辨率和丰度估计，从而避免了两阶段模型中常见的单向误差传播。我们通过模拟和使用阿拉斯加冰川湾海獭航拍图像的案例研究来说明我们的方法。本文随附的补充材料已在线发布 从而避免在两阶段模型中常见的单向误差传播。我们通过模拟和使用阿拉斯加冰川湾海獭航拍图像的案例研究来说明我们的方法。本文随附的补充材料已在线发布 从而避免在两阶段模型中常见的单向误差传播。我们通过模拟和使用阿拉斯加冰川湾海獭航拍图像的案例研究来说明我们的方法。本文随附的补充材料已在线发布