Weeds are a crucial threat to agriculture, and in order to preserve crop productivity, spreading agrochemicals is a common practice with a potential negative impact on the environment. Methods that can support intelligent application are needed. Therefore, identification and mapping is a critical step in performing site-specific weed management. Unmanned aerial vehicle (UAV) data streams are considered the best for weed detection due to the high resolution and flexibility of data acquisition and the spatial explicit dimensions of imagery. However, with the existence of unstructured crop conditions and the high biological variation of weeds, it remains a difficult challenge to generate accurate weed recognition and detection models. Two critical barriers to tackling this challenge are related to (1) a lack of case-specific, large, and comprehensive weed UAV image datasets for the crop of interest, (2) defining the most appropriate computer vision (CV) weed detection models to assess the operationality of detection approaches in real case conditions. Deep Learning (DL) algorithms, appropriately trained to deal with the real case complexity of UAV data in agriculture, can provide valid alternative solutions with respect to standard CV approaches for an accurate weed recognition model. In this framework, this paper first introduces a new weed and crop dataset named Chicory Plant (CP) and then tests state-of-the-art DL algorithms for object detection. A total of 12,113 bounding box annotations were generated to identify weed targets (Mercurialis annua) from more than 3000 RGB images of chicory plantations, collected using a UAV system at various stages of crop and weed growth. Deep weed object detection was conducted by testing the most recent You Only Look Once version 7 (YOLOv7) on both the CP and publicly available datasets (Lincoln beet (LB)), for which a previous version of YOLO was used to map weeds and crops. The YOLOv7 results obtained for the CP dataset were encouraging, outperforming the other YOLO variants by producing value metrics of 56.6%, 62.1%, and 61.3% for the mAP@0.5 scores, recall, and precision, respectively. Furthermore, the YOLOv7 model applied to the LB dataset surpassed the existing published results by increasing the mAP@0.5 scores from 51% to 61%, 67.5% to 74.1%, and 34.6% to 48% for the total mAP, mAP for weeds, and mAP for sugar beets, respectively. This study illustrates the potential of the YOLOv7 model for weed detection but remarks on the fundamental needs of large-scale, annotated weed datasets to develop and evaluate models in real-case field circumstances.

During the past decade, imagery data acquired from unmanned aerial vehicles (UAVs), thanks to their high spatial, spectral, and temporal resolutions, have attracted increasing attention for discriminating healthy from diseased plants and monitoring the progress of such plant diseases in fields. Despite the well-documented usage of UAV-based hyperspectral remote sensing for discriminating healthy and diseased plant areas, employing red-green-blue (RGB) imagery for a similar purpose has yet to be fully investigated. This study aims at evaluating UAV-based RGB imagery to discriminate healthy plants from those infected by stripe and wheat leaf rusts in winter wheat (Triticum aestivum L.), with a focus on implementing an expert system to assist growers in improved disease management. RGB images were acquired at four representative wheat-producing sites in the Grand Duchy of Luxembourg. Diseased leaf areas were determined based on the digital numbers (DNs) of green and red spectral bands for wheat stripe rust (WSR), and the combination of DNs of green, red, and blue spectral bands for wheat leaf rust (WLR). WSR and WLR caused alterations in the typical reflectance spectra of wheat plants between the green and red spectral channels. Overall, good agreements between UAV-based estimates and observations were found for canopy cover, WSR, and WLR severities, with statistically significant correlations (p-value (Kendall) < 0.0001). Correlation coefficients were 0.92, 0.96, and 0.86 for WSR severity, WLR severity, and canopy cover, respectively. While the estimation of canopy cover was most often less accurate (correlation coefficients < 0.20), WSR and WLR infected leaf areas were identified satisfactorily using the RGB imagery-derived indices during the critical period (i.e., stem elongation and booting stages) for efficacious fungicide application, while disease severities were also quantified accurately over the same period. Using such a UAV-based RGB imagery method for monitoring fungal foliar diseases throughout the cropping season can help to identify any new disease outbreak and efficaciously control its spread.

Remote sensing data play a crucial role in monitoring crop dynamics in the context of precision agriculture by characterizing the spatial and temporal variability of crop traits. At present there is special interest in assessing the long-term impacts of biochar in agro-ecosystems. Despite the growing body of literature on monitoring the potential biochar effects on harvested crop yield and aboveground productivity, studies focusing on the detailed crop performance as a consequence of long-term biochar enrichment are still lacking. The primary objective of this research was to evaluate crop performance based on high-resolution unmanned aerial vehicle (UAV) imagery considering both crop growth and health through RGB and multispectral analysis, respectively. More specifically, this approach allowed monitoring of century-old biochar impacts on winter wheat crop performance. Seven Red-Green-Blue (RGB) and six multispectral flights were executed over 11 century-old biochar patches of a cultivated field. UAV-based RGB imagery exhibited a significant positive impact of century-old biochar on the evolution of winter wheat canopy cover (p-value = 0.00007). Multispectral optimized soil adjusted vegetation index indicated a better crop development over the century-old biochar plots at the beginning of the season (p-values < 0.01), while there was no impact towards the end of the season. Plant height, derived from the RGB imagery, was slightly higher for century-old biochar plots. Crop health maps were computed based on principal component analysis and k-means clustering. To our knowledge, this is the first attempt to quantify century-old biochar effects on crop performance during the entire growing period using remotely sensed data. Ground-based measurements illustrated a significant positive impact of century-old biochar on crop growth stages (p-value of 0.01265), whereas the harvested crop yield was not affected. Multispectral simplified canopy chlorophyll content index and normalized difference red edge index were found to be good linear estimators of harvested crop yield (p-value_(Kendall) of 0.001 and 0.0008, respectively). The present research highlights that other factors (e.g., inherent pedological variations) are of higher importance than the presence of century-old biochar in determining crop health and yield variability.

It can be challenging to fuse remotely-sensed images with large differences in spatial resolutions. In this paper, we used additive wavelet transform (AWT) to fuse Landsat-8 (30 m) and unmanned aerial vehicle (UAV) images (7 cm and 3.7 cm for thermal and multispectral images, respectively) as one of the primary studies. AWT image fusion generated sharpened Landsat-8 (L-8) images which were significantly correlated with coarse resolution images, while also well preserving the spatial details. Surface albedo (α₀), normalized difference vegetation index (NDVI), and surface temperature (ST) were computed from multispectral and thermal sensors on board of UAV and L-8 platforms. High-resolution UAV and AWT sharpened L-8 images were then used in ETLook model to estimate evapotranspiration (ET) across an agricultural farm enriched with century-old biochar. High spatio-temporal analysis demonstrated a significant decrease in α₀ across the biochar patches during the early development stages of winter wheat. Moreover, biochar significantly stimulated the development of wheat canopies towards the middle of the cropping season. There were however no impacts at the end of the season due to dense wheat canopies covering the aggravated dark colour soil across the biochar patches. ST was not affected by biochar either at the beginning or towards the end of the season. Neither was there any impact of biochar on actual ET over the season. Our approach can help to develop robust techniques for fusion of UAV and satellite images in light of climate-smart agriculture, and is also applicable to other farms with any specific precision agricultural treatments. GRAPHICAL ABSTRACT

At present, it is critical to accurately monitor wheat crops to help decision-making processes in precision agriculture. This research aims to retrieve various wheat crop traits from hyperspectral data using machine learning regression algorithms (MLRAs) and dimensionality reduction (DR) techniques. This experiment was conducted in an agricultural field in Arborea, Oristano-Sardinia, Italy, with different factors such as cultivars, N-treatments, and soil ploughing conditions. Hyperspectral data were acquired on the ground using a full-range Spectral Evolution spectrometer (350–2500 nm). Four DR techniques, including (i) variable influence on projection (VIP), (ii) principal component analysis (PCA), (iii) vegetation indices (VIs), and (iv) spectroscopic feature (SF) calculation, were undertaken to reduce the dimension of the hyperspectral data while maintaining the information content. We used five MLRA models, including (i) partial least squares regression (PLSR), (ii) random forest (RF), (iii) support vector regression (SVR), (iv) Gaussian process regression (GPR), and (v) neural network (NN), to retrieve wheat traits at either leaf and canopy levels. The studied traits were leaf area index (LAI), leaf and canopy water content (LWC and CWC), leaf and canopy chlorophyll content (LCC and CCC), and leaf and canopy nitrogen content (LNC and CNC). MLRA models were able to accurately retrieve wheat traits at the canopy level with PLSR and NN indicating the highest modelling performance. On the contrary, MLRA models indicated less accurate retrievals of the leaf-level traits. DR techniques were found to notably improve the retrieval accuracy of crop traits. Furthermore, the generated models were re-calibrated using soil spectra and then transferred to an airborne dataset collected using a CASI-SASI hyperspectral sensor, allowing the estimation of wheat traits across the entire field. The predicted crop trait maps illustrated consistent patterns while also preserving the real-field characteristics well. Lastly, a statistical paired t-test was undertaken to conduct a proof of concept of wheat phenotyping analysis considering the different agricultural variables across the study site. N-treatment caused significant differences in wheat crop traits in many instances, whereas the observed differences were less pronounced between the cultivars. No particular impact of soil ploughing conditions on wheat crop characteristics was found. Using such combinations of MLRA and DR techniques based on hyperspectral data can help to effectively monitor crop traits throughout the cropping seasons and can also be readily applied to other agricultural settings to help both precision farming applications and the implementation of high-throughput phenotyping solutions.