An important consideration for UAV-based (unmanned aerial vehicle) object detection in the natural environment is vegetation height and foliar cover, which can visually obscure the items a machine learning model is trained to detect. Hence, the accuracy of aerial detection of objects such as surface landmines and UXO (unexploded ordnance) is highly dependent on the height and density of vegetation in a given area. In this study, we develop a model that estimates the detection accuracy (recall) of a YOLOv8 object’s detection implementation as a function of occlusion due to vegetation coverage. To solve this function, we developed an algorithm to extract vegetation height and coverage of the UAV imagery from a digital surface model generated using structure-from-motion (SfM) photogrammetry. We find the relationship between recall and percent occlusion is well modeled by a sigmoid function using the PFM-1 landmine test case. Applying the sigmoid recall-occlusion relationship in conjunction with our vegetation cover algorithm to solve for percent occlusion, we mapped the uncertainty in detection rate due to vegetation in UAV-based SfM orthomosaics in eight different minefield environments. This methodology and model have significant implications for determining the optimal location and time of year for UAV-based object detection tasks and quantifying the uncertainty of deep learning object detection models in the natural environment.

Most lava flows carry bubbles and crystals in suspension. From earlier works, it is known that spherical bubbles increase the effective viscosity while bubbles deformed by rapid flow decrease it. Changes in the spatial distribution of bubbles can lead to variable rheology and flow localization and thus modify the resulting lava flow structure and morphology. To understand the roles of bubble and solid phase crystal distributions, we conducted a series of analog experiments of high bubble fraction suspensions. We poured the analog lava on an inclined slope, observed its shape, calculated the velocity field, and monitored its local thickness. A region of localized rapid flow and low vesicularity, whose thickness is thinner than the surrounding area, develops at the center of the bubbly flows. These features suggest that the locally higher liquid fraction decreases the effective viscosity, increases the fluid density, and accelerates the flow. We also found that a halted particle‐bearing bubbly flow can resume flowing. We interpret this to result from the upward vertical separation of bubbles, which generates a liquid‐rich layer at the bottom of the flow. In our experiment, bubbles are basically spherical and decrease the flow velocity, while our estimate suggests that bubbles in natural lava flows could increase or decrease flow velocity. Downstream decreases in flow velocity stops the bubble deformation and can cause a sudden increase of effective viscosity. The vertical segregation of the liquid phase at the slowed flow front may be a way to generate a cavernous shelly paho’eho’e.

Existing frameworks for classifying the fidelity for object detection tasks do not consider false positive likelihood and object uniqueness. Inspired by the Detection, Recognition, Identification (DRI) framework proposed by Johnson 1958, we propose a new modified framework that defines three categories as visible anomaly, identifiable anomaly, and unique identifiable anomaly (AIU) as determined by human interpretation of imagery or geophysical data. These categories are designed to better capture false positive rates and emphasize the importance of identifying unique versus non-unique targets compared to the DRI Index. We then analyze visual, thermal, and multispectral UAV imagery collected over a seeded minefield and apply the AIU Index for the landmine detection use-case. We find that RGB imagery provided the most value per pixel, achieving a 100% identifiable anomaly rate at 125 pixels on target, and the highest unique target classification compared to thermal and multispectral imaging for the detection and identification of surface landmines and UXO. We also investigate how the AIU Index can be applied to machine learning for the selection of training data and informing the required action to take after object detection bounding boxes are predicted. Overall, the anomaly, identifiable anomaly, and unique identifiable anomaly index prescribes essential context for false-positive-sensitive or resolution-poor object detection tasks with applications in modality comparison, machine learning, and remote sensing data acquisition.

Around the world, approximately 110 million explosive ordinances (EO), including remnant antipersonnel landmines, antivehicle/tank mines, unexploded cluster munitions, improvised explosive devices (IEDs), explosive remnants of war (ERW) like unexploded ordnances (UXOs) and abandoned explosive ordinances (AXOs), remain untouched and undetected in post-conflict nations. Current geophysical EO detection methods include ground-based electromagnetic induction (EMI), which has proven to be costly, dangerous, and time consuming. Recent hardware advances have allowed for drastic reductions in the size, weight, power, and cost (SWaP-C) of miniaturized geophysical sensors. Sensors like hyperspectral imaging (HSI) systems can now be successfully mounted to small unpiloted aerial vehicles (UAS). In this proof-of-concept study, we combine UAV-based HSI in an attempt to develop and calibrate a safer, faster, and more cost efficient method of remote detection of surficial EO. We collected data with a Corning microHSI 410 selectable hyperspectral airborne remote sensing kit (SHARK) attached to a DJI Matrice 600 with a Ronin-MX gimbal in simulated minefields over several different environmental settings. We emplaced multiple types of ERWs, including anti-personnel mines, anti-vehicle mines, rockets, missiles, and grenades. Significantly, several of these are currently being used in the Russo-Ukrainian War, including: TM-62M and TM-62P antivehicle mines; 40 mm submunitions; OZM-72, PFM-1, PMN, and PMN-4 antipersonnel mines; 82 mm mortars; multiple types of cluster bombs; 122 mm GRAD rockets; PG-7 rocket propelled grenades; and several fuzes. We found that HSI proved to successfully image a variety of EO, from large PG-7 rocket propelled grenades (RPGs) to the small, notoriously difficult to detect, minimal metal PFM1 anti-personnel landmine. This study allowed us to better understand the strengths and weaknesses of UAV-based HSI to detect and spectrally quantify ERW of various sizes and material compositions. Our analysis demonstrated that unique spectral profiles and derivative data products can be created to identify multiple EO in relatively high resolution hyperspectral images and in variable environments. Future research will integrate HSI data as additional channels to the convolutional neural network (CNN) algorithm our group has been developing.

Reliable and scalable landmine detection technologies are essential for humanitarian mine action (HMA), yet standardized benchmarks for Unmanned Aerial Vehicle (UAV)-based sensing in operationally relevant environments remain limited. This study presents a comprehensive evaluation of 34 multimodal datasets acquired over a standardized seeded test site for landmine and unexploded ordnance detection. Nine sensing modalities, including RGB, thermal, multispectral, hyperspectral, LiDAR, and Synthetic Aperture Radar (SAR), are evaluated using the Anomaly, Identifiable Anomaly, Unique Identifiable Anomaly (AIU) index to establish a unified framework for quantifying detection fidelity. Results indicate that RGB imagery achieves the highest surface detection rate (94.8%), with 45.4% of targets classified as uniquely identifiable, reducing false-positive risk. For sub-surface detection, handheld electromagnetic induction (EMI) and magnetometry exceed 95% detection for ferrous items but fall below 10% for plastic ordnance. Ground-penetrating radar (GPR) is the only modality capable of detecting buried plastic targets (55.6% for cart-based systems), whereas UAV-mounted GPR remains limited (18.2%) at current operational flight heights. Based on the comparative analysis, we discuss the gaps in current detection capabilities, compare false positive rates across modalities, and perform a cost-benefit analysis fitting contamination scenario with the most cost effective detection method. All datasets are publicly released (https://zenodo.org/records/19100554) along with an interactive web-map (https://staging.dmsamfvv1bhon.amplifyapp.com/) to support reproducible benchmarking and cross-modality comparison in UAV-enabled explosive hazard detection.