Kobus Barnard

Charcoal rot of sorghum (CRS) is a significant disease affecting sorghum crops, with limited genetic resistance available. The causative agent, Macrophomina phaseolina (Tassi) Goid, is a highly destructive fungal pathogen that targets over 500 plant species globally, including essential staple crops. Utilizing field image data for precise detection and quantification of CRS could greatly assist in the prompt identification and management of affected fields and thereby reduce yield losses. The objective of this work was to implement various machine learning algorithms to evaluate their ability to accurately detect and quantify CRS in red‐green‐blue images of sorghum plants exhibiting symptoms of infection. EfficientNet‐B3 and a fully convolutional network emerged as the top‐performing models for image classification and segmentation tasks, respectively. Among the classification models evaluated, EfficientNet‐B3 demonstrated superior performance, achieving an accuracy of 86.97%, a recall rate of 0.71, and an F1 score of 0.73. Of the segmentation models tested, FCN proved to be the most effective, exhibiting a validation accuracy of 97.76%, a recall rate of 0.68, and an F1 score of 0.66. As the size of the image patches increased, both models’ validation scores increased linearly, and their inference time decreased exponentially. This trend could be attributed to larger patches containing more information, improving model performance, and fewer patches reducing the computational load, thus decreasing inference time. The models, in addition to being immediately useful for breeders and growers of sorghum, advance the domain of automated plant phenotyping and may serve as a foundation for drone‐based or other automated field phenotyping efforts. Additionally, the models presented herein can be accessed through a web‐based application where users can easily analyze their own images.

Charcoal rot of sorghum (CRS) is a significant disease affecting sorghum crops, with limited genetic resistance available. The causative agent, Macrophomina phaseolina (Tassi) Goid, is a highly destructive fungal pathogen that targets over 500 plant species globally, including essential staple crops. Utilizing field image data for precise detection and quantification of CRS could greatly assist in the prompt identification and management of affected fields and thereby reduce yield losses. The objective of this work was to implement various machine learning algorithms to evaluate their ability to accurately detect and quantify CRS in red-green-blue (RGB) images of sorghum plants exhibiting symptoms of infection. EfficientNet-B3 and a fully convolutional network (FCN) emerged as the top-performing models for image classification and segmentation tasks, respectively. Among the classification models evaluated, EfficientNet-B3 demonstrated superior performance, achieving an accuracy of 86.97%, a recall rate of 0.71, and an F1 score of 0.73. Of the segmentation models tested, FCN proved to be the most effective, exhibiting a validation accuracy of 97.76%, a recall rate of 0.68, and an F1 score of 0.66. As the size of the image patches increased, both models’ validation scores increased linearly, and their processing time decreased exponentially. The models, in addition to being immediately useful for breeders and growers of sorghum, advance the domain of automated plant phenotyping and may serve as a base for drone-based or other automated field phenotyping efforts. Additionally, the models presented herein can be accessed through a web-based application where users can easily analyze their own images. Core ideas: Automated phenotyping tools are required for the efficient detection and quantification of charcoal rot of sorghum. Classification and segmentation models can distinguish between concurrent plant stresses with similar symptoms. Larger image patch sizes generally improve model performance and reduce processing time.

The Legacy Survey of Space and Time (LSST) at the Vera C. Rubin Observatory will discover tens of thousands of extragalactic transients each night. The high volume of alerts demands immediate classification of transient types in order to prioritize observational follow-ups before events fade away. We use host galaxy features to classify transients, thereby providing classification upon discovery. In contrast to past work that focused on distinguishing Type Ia and core-collapse supernovae (SNe) using host galaxy features that are not always accessible (e.g., morphology), we determine the relative likelihood across 12 transient classes based on only 19 host apparent magnitudes and colors from 10 optical and IR photometric bands. We develop both binary and multiclass classifiers, using kernel density estimation to estimate the underlying distribution of host galaxy properties for each transient class. Even in this pilot study, and ignoring relative differences in transient class frequencies, we distinguish eight transient classes at purities significantly above the 8.3% baseline (based on a classifier that assigns labels uniformly and at random): tidal disruption events (TDEs; 48% ± 27%, where ± indicates the 95% confidence limit), SNe Ia-91bg (32% ± 18%), SNe Ia-91T (23% ± 11%), SNe Ib (23% ± 13%), SNe II (17% ± 2%), SNe IIn (17% ± 6%), SNe II P (16% ± 4%), and SNe Ia (10% ± 1%). We demonstrate that our model is applicable to LSST and estimate that our approach can accurately classify 59% of LSST alerts expected each year for SNe Ia, Ia-91bg, II, Ibc, SLSN-I, and TDEs. Our code and data set are publicly available.

Understanding the preferences of transient types for host galaxies with certain characteristics is key to studies of transient physics and galaxy evolution, as well as to transient identification and classification in the LSST era. Here we describe a value-added database of extragalactic transients—supernovae, tidal disruption events, gamma-ray bursts, and other rare events—and their host galaxy properties. Based on reported coordinates, redshifts, and host galaxies (if known) of events, we cross-identify their host galaxies or most likely host candidates in various value-added or survey catalogs, and compile the existing photometric, spectroscopic, and derived physical properties of the host galaxies in these catalogs. This new database covers photometric measurements from the far-ultraviolet to mid-infrared. Spectroscopic measurements and derived physical properties are also available for a smaller subset of hosts. For our 36,333 unique events, we have cross-identified 13,753 host galaxies using host names, plus 4480 using host coordinates. Besides those with known hosts, there are 18,100 transients with newly identified host candidates. This large database will allow explorations of the connections of transients to their hosts, including a path toward transient alert filtering and probabilistic classification based on host properties.

We address fast image stitching for large image collections while being robust to drift due to chaining transformations and minimal overlap between images. We focus on scientific applications where ground-truth accuracy is far more important than visual appearance or projection error, which can be misleading. For common large-scale image stitching use cases, transformations between images are often restricted to similarity or translation. When homography is used in these cases, the odds of being trapped in a poor local minimum and producing unnatural results increases. Thus, for transformations up to affine, we cast stitching as minimizing reprojection error globally using linear least-squares with a few, simple constraints. For homography, we observe that the global affine solution provides better initialization for bundle adjustment compared to an alternative that initializes with a homography-based scaffolding and at lower computational cost. We evaluate our methods on a very large translation dataset with limited overlap as well as four drone datasets. We show that our approach is better compared to alternative methods such as MGRAPH in terms of computational cost, scaling to large numbers of images, and robustness to drift. We also contribute ground-truth datasets for this endeavor.

To mitigate the issue of minimal intrinsic features for pure data-driven methods, in this article, we propose a novel model-driven deep network for infrared small target detection, which combines discriminative networks and conventional model-driven methods to make use of both labeled data and the domain knowledge. By designing a feature map cyclic shift scheme, we modularize a conventional local contrast measure method as a depthwise parameterless nonlinear feature refinement layer in an end-to-end network, which encodes relatively long-range contextual interactions with clear physical interpretability. To highlight and preserve the small target features, we also exploit a bottom-up attentional modulation integrating the smaller scale subtle details of low-level features into high-level features of deeper layers. We conduct detailed ablation studies with varying network depths to empirically verify the effectiveness and efficiency of the design of each component in our network architecture. We also compare the performance of our network against other model-driven methods and deep networks on the open SIRST data set as well. The results suggest that our network yields a performance boost over its competitors.

Feature fusion, the combination of features from different layers or branches, is an omnipresent part of modern network architectures. It is often implemented via simple operations, such as summation or concatenation, but this might not be the best choice. In this work, we propose a uniform and general scheme, namely attentional feature fusion, which is applicable for most common scenarios, including feature fusion induced by short and long skip connections as well as within Inception layers. To better fuse features of inconsistent semantics and scales, we propose a multi-scale channel attention module, which addresses issues that arise when fusing features given at different scales. We also demonstrate that the initial integration of feature maps can become a bottleneck and that this issue can be alleviated by adding another level of attention, which we refer to as iterative attentional feature fusion. With fewer layers or parameters, our models outperform state-of-the-art networks on both CIFAR-100 and ImageNet datasets, which suggests that more sophisticated attention mechanisms for feature fusion hold great potential to consistently yield better results compared to their direct counterparts. Our codes and trained models are available online.

Activation functions and attention mechanisms are typically treated as having different purposes and have evolved differently. However, both concepts can be formulated as a non-linear gating function. Inspired by their similarity, we propose a novel type of activation units called attentional activation (ATAC) units as a unification of activation functions and attention mechanisms. In particular, we propose a local channel attention module for the simultaneous non-linear activation and element-wise feature refinement, which locally aggregates point-wise cross-channel feature contexts. By replacing the well-known rectified linear units by such ATAC units in convolutional networks, we can construct fully attentional networks that perform significantly better with a modest number of additional parameters. We conducted detailed ablation studies on the ATAC units using several host networks with varying network depths to empirically verify the effectiveness and efficiency of the units. Furthermore, we compared the performance of the ATAC units against existing activation functions as well as other attention mechanisms on the CIFAR-10, CIFAR-100, and ImageNet datasets. Our experimental results show that networks constructed with the proposed ATAC units generally yield performance gains over their competitors given a comparable number of parameters.

Glacial science could benefit tremendously from autonomous robots, but previous glacial robots have had perception issues in these colorless and featureless environments, specifically with visual feature extraction. This translates to failures in visual odometry and visual navigation. Glaciologists use near-infrared imagery to reveal the underlying heterogeneous spatial structure of snow and ice, and we theorize that this hidden near-infrared structure could produce more and higher quality features than available in visible light. We took a custom camera rig to Igloo Cave at Mt. St. Helens to test our theory. The camera rig contains two identical machine vision cameras, one which was outfitted with multiple filters to see only near-infrared light. We extracted features from short video clips taken inside Igloo Cave at Mt. St. Helens, using three popular feature extractors (FAST, SIFT, and SURF). We quantified the number of features and their quality for visual navigation by comparing the resulting orientation estimates to ground truth. Our main contribution is the use of NIR longpass filters to improve the quantity and quality of visual features in icy terrain, irrespective of the feature extractor used.