The digital age is profoundly changing how data is collected, stored and analyzed in the ecological sciences. Where scientists would previously resort to in-situ observations of nature, they are now increasingly relying on digital recording devices to collect large data-sets. Collected data typically requires annotation in order to yield the desired data-products, but unlike collection, annotation often requires expert skills, which are in short supply. As a consequence, a manual-annotation bottleneck has developed in many disciplines between collected data and desired data-products. Concurrently, significant advances have been made in automated identification of common objects & scenes using machine learning and computer vision methods. This thesis builds on these advances in order to address the manual- annotation bottleneck in coral reef surveys. The focus on reef surveys was established due to the severity of the manual-annotation bottleneck, and the critical importance of large-scale reef surveys in understanding the contemporary global coral reef crisis. Towards this goal, we investigate several aspects of the data collection and annotation work. First, methods for automated annotation of coral reef survey images are developed and compared to human expert annotators. Random sampling designs and semi- automated frameworks are proposed which enable deployment of automated annotation without sacrificing the estimation quality of derived data-products (e.g. percent coral cover). Such methods are appealing as they fuse the high accuracy of expert annotators with the speed and low cost of automated annotators. Further, a multi-channel annotation algorithm, which utilizes broad-spectrum fluorescence images in conjunction with standard reflectance images, is proposed and shown to significantly reduce the error-rate compared to using reflectance images only. Finally, the developed methods are implemented in a web-portal, CoralNet (coralnet.ucsd.edu), and made publicly available. At the time of writing, over 80,000 images have been uploaded and annotated by over 200 users, including research labs and government agencies

Global climate change and other anthropogenic stressors have heightened the need to rapidly characterize ecological changes in marine benthic communities across large scales. Digital photography enables rapid collection of survey images to meet this need, but the subsequent image annotation is typically a time consuming, manual task. We investigated the feasibility of using automated point-annotation to expedite cover estimation of the 17 dominant benthic categories from survey-images captured at four Pacific coral reefs. Inter- and intra- annotator variability among six human experts was quantified and compared to semi- and fully- automated annotation methods, which are made available at coralnet.ucsd.edu. Our results indicate high expert agreement for identification of coral genera, but lower agreement for algal functional groups, in particular between turf algae and crustose coralline algae. This indicates the need for unequivocal definitions of algal groups, careful training of multiple annotators, and enhanced imaging technology. Semi-automated annotation, where 50% of the annotation decisions were performed automatically, yielded cover estimate errors comparable to those of the human experts. Furthermore, fully-automated annotation yielded rapid, unbiased cover estimates but with increased variance. These results show that automated annotation can increase spatial coverage and decrease time and financial outlay for image-based reef surveys.