Toward the remote monitoring of armed conflicts.

The war in Ukraine has pushed the role of satellite imagery in armed conflicts into the spotlight. For a long time, satellite images were primarily used for military and intelligence purposes, but today they permeate every aspect of armed conflicts. Their importance in influencing the course of armed conflicts will further grow as progress in deep learning makes automated analysis progressively possible. This article assesses the state of the research working toward the remote monitoring of armed conflicts and highlights opportunities to increase the positive societal impact of future research efforts. First, we map the existing literature, categorizing studies in terms of conflict events that are covered, conflict context and scope, techniques, and types of satellite imagery used to identify conflict events. Second, we discuss how these choices affect opportunities to develop applications for human rights, humanitarian, and peacekeeping actors. Third, we provide an outlook, assessing promising paths forward. While much focus has been on high spatial resolution imagery, we demonstrate why research on freely available satellite images with moderate spatial but high temporal resolution can lead to more scalable and transferable options. We argue that research on such images should be prioritized, as it will have a greater positive impact on society, and we discuss what types of applications may soon become feasible through such research. We call for concerted efforts to compile a large dataset of nonsensitive conflict events to accelerate research toward the remote monitoring of armed conflicts and for interdisciplinary collaboration to ensure conflict-sensitive monitoring solutions.

Cocoa plantations are associated with deforestation in Côte d'Ivoire and Ghana.

Côte d'Ivoire and Ghana, the world's largest producers of cocoa, account for two thirds of the global cocoa production. In both countries, cocoa is the primary perennial crop, providing income to almost two million farmers. Yet precise maps of the area planted with cocoa are missing, hindering accurate quantification of expansion in protected areas, production and yields and limiting information available for improved sustainability governance. Here we combine cocoa plantation data with publicly available satellite imagery in a deep learning framework and create high-resolution maps of cocoa plantations for both countries, validated in situ. Our results suggest that cocoa cultivation is an underlying driver of over 37% of forest loss in protected areas in Côte d'Ivoire and over 13% in Ghana, and that official reports substantially underestimate the planted area (up to 40% in Ghana). These maps serve as a crucial building block to advance our understanding of conservation and economic development in cocoa-producing regions.

The Herbarium 2021 Half-Earth Challenge Dataset and Machine Learning Competition.

Herbarium sheets present a unique view of the world's botanical history, evolution, and biodiversity. This makes them an all-important data source for botanical research. With the increased digitization of herbaria worldwide and advances in the domain of fine-grained visual classification which can facilitate automatic identification of herbarium specimen images, there are many opportunities for supporting and expanding research in this field. However, existing datasets are either too small, or not diverse enough, in terms of represented taxa, geographic distribution, and imaging protocols. Furthermore, aggregating datasets is difficult as taxa are recognized under a multitude of names and must be aligned to a common reference. We introduce the Herbarium 2021 Half-Earth dataset: the largest and most diverse dataset of herbarium specimen images, to date, for automatic taxon recognition. We also present the results of the Herbarium 2021 Half-Earth challenge, a competition that was part of the Eighth Workshop on Fine-Grained Visual Categorization (FGVC8) and hosted by Kaggle to encourage the development of models to automatically identify taxa from herbarium sheet images.

Fusion of building information and range imaging for autonomous location estimation in indoor environments.

We present a novel approach for autonomous location estimation and navigation in indoor environments using range images and prior scene knowledge from a GIS database (CityGML). What makes this task challenging is the arbitrary relative spatial relation between GIS and Time-of-Flight (ToF) range camera further complicated by a markerless configuration. We propose to estimate the camera's pose solely based on matching of GIS objects and their detected location in image sequences. We develop a coarse-to-fine matching strategy that is able to match point clouds without any initial parameters. Experiments with a state-of-the-art ToF point cloud show that our proposed method delivers an absolute camera position with decimeter accuracy, which is sufficient for many real-world applications (e.g., collision avoidance).