Search

Early detection of drought stress is essential for preventing permanent plant damage and minimizing yield loss. This study utilized hyperspectral imaging at the leaf level to visualize drought stress in safflower plants (Carthamus tinctorius L.). Three safflower genotypes, Palenus, A82, and IL-111, were cultivated under three irrigation levels. Stress conditions were simulated by depleting 50%, 70%, and 90% of soil water content, representing unstressed (US), mild stress (MS), and severe stress (SS) conditions, respectively. Hyperspectral images of leaf samples were captured before any visible signs of water scarcity emerged. Classification analysis was performed using the full mean spectral data with partial least squares discriminant analysis, soft independent modeling of class analogy (SIMCA), support vector machines, and artificial neural network (ANN) classifiers. Feature selection methods were applied to extract the most informative wavebands, and ANN was used to build predictive models. Spatial analysis involved pixel-wise classification using both unsupervised (k-means clustering) and supervised (best classifiers) approaches. ANN outperformed other classifiers using the entire spectral data, effectively distinguishing US, MS, and SS classes in the Palenus, A82, and IL-111 genotypes, achieving F1-scores of 92.22%, 96.01%, and 96.47%, respectively. Among the feature selection methods, SIMCA-based features excelled in monitoring stress conditions in the Palenus and A82 genotypes. In supervised spatial analysis, ANN models clearly depicted the progression of stress in leaves across different genotypes. This study demonstrates the potential of hyperspectral imaging to differentiate various levels of drought stress in safflower, an important oilseed crop.

Monitoring of milk composition can support several dimensions of dairy management such as identification of the health status of individual dairy cows and the safeguarding of dairy quality. The quantification of milk composition has been traditionally executed employing destructive chemical or laboratory Fourier-transform infrared (FTIR) spectroscopy analyses which can incur high costs and prolonged waiting times for continuous monitoring. Therefore, modern technology for milk composition quantification relies on non-destructive near-infrared (NIR) spectroscopy which is not invasive and can be performed on-farm, in real-time.The current dataset contains NIR spectral measurements in transmittance mode in the wavelength range from 960 nm to 1690 nm of 1224 individual raw milk samples, collected on-farm over an eight-week span in 2017, at the experimental dairy farm of the province of Antwerp, ‘Hooibeekhoeve’ (Geel, Belgium). For these spectral measurements, laboratory reference values corresponding to the three main components of raw milk (fat, protein and lactose), urea and somatic cell count (SCC) are included.This data has been used to build multivariate calibration models to predict the three milk compounds, as well as develop strategies to monitor the prediction performance of the calibration models.

The spotted wing Drosophila (SWD), Drosophila suzukii, is a significant invasive pest of berries and soft-skinned fruits that causes major economic losses in fruit production worldwide. Automatic identification and monitoring strategies would allow to detect the emergence of this pest in an early stage and minimize its impact. The small size of Drosophila suzukii and similar flying insects makes it difficult to identify them using camera systems. Therefore, an optical sensor recording wingbeats was investigated in this study. We trained convolutional neural network (CNN) classifiers to distinguish D. suzukii insects from one of their closest relatives, Drosophila Melanogaster, based on their wingbeat patterns recorded by the optical sensor. Apart from the original wingbeat time signals, we modeled their frequency (power spectral density) and time-frequency (spectrogram) representations. A strict validation procedure was followed to estimate the models’ performance in field-conditions. First, we validated each model on wingbeat data that was collected under the same conditions using different insect populations to train and test them. Next, we evaluated their robustness on a second independent dataset which was acquired under more variable environmental conditions. The best performing model, named “InceptionFly,” was trained on wingbeat time signals. It was able to discriminate between our two target insects with a balanced accuracy of 92.1% on the test set and 91.7% on the second independent dataset. This paves the way towards early, automated detection of D. suzukii infestation in fruit orchards.

In the egg industry, fast and highly reliable quality measurements are crucial. This study presents a novel method based on Hertz contact theory that allows for non-destructive determination of eggshell strength. The goal of the study was to evaluate the material strength (Young’s Modulus) and structural strength (stiffness) of eggshells. To this end, an experimental setup was constructed to measure the collision of an eggshell with a small steel ball, which was recorded using a laser vibrometer. The study analyzed a sample of 120 eggs and found a correlation of 0.85 between the traditional static stiffness measured during quasi-static compression tests and the stiffness obtained from the Hertz contact theory. The results show that Hertz contact theory is valid for small steel spheres impacting eggshells, while a sensitivity analysis indicated that the most important factor in determining the strength of the eggshell is the contact duration between the egg and the impactor. These results open up the possibility of grading eggs based on their shell strength in a non-destructive manner.

Eggshell strength is a critical quality factor for consumption eggs as it affects the probability of breakage in practice. In this study, a fast and low-cost methodology for the non-destructive determination of eggshell strength is presented. The method utilized a small steel ball to impact the egg and a microphone to analyse the impact characteristics. Hertz contact theory was applied to relate the measured impact characteristics to the local stiffness of the eggshell. Therefore, a total of 150 eggs were studied on which eight consecutive measurements per egg were taken around the equator at equidistant places. The results showed a strong correlation of 0.93 between the traditional static stiffness measured during quasi-static compression tests and the average stiffness obtained from the new methodology. This paves the way towards fast, low-cost and non-destructive in-line shell strength measurements to reduce the number of cracked eggs reaching the consumer.

Leek farmers tend to apply too much nitrogen fertilizer as its cost is relatively low compared to the gross value of leek. Recently, several studies have shown that proximal sensing technologies could accurately monitor the crop nitrogen content and biomass. However, their implementation is impeded by practical limitations and the limited area they can cover. UAV-based monitoring might alleviate these issues. Studies on UAV-based vegetable crop monitoring are still limited. Because of the economic importance and environmental impact of leeks in Flanders, this study aimed to investigate the ability of UAV-based multispectral imaging to accurately monitor leek nitrogen uptake and dry biomass across multiple fields and seasons. Different modelling approaches were tested using twelve spectral VIs and the interquartile range of each of these VIs within the experimental plots as predictors. In a leave-one-flight out cross-validation (LOF-CV), leek dry biomass (DBM) was most accurately predicted using a lasso regression model (RMSEct = 6.60 g plant−1, R2= 0.90). Leek N-uptake was predicted most accurately by a simple linear regression model based on the red wide dynamic range (RWDRVI) (RMSEct = 0.22 gN plant−1, R2 = 0.85). The results showed that randomized Kfold-CV is an undesirable approach. It resulted in more consistent and lower RMSE values during model training and selection, but worse performance on new data. This would be due to information leakage of flight-specific conditions in the validation data split. However, the model predictions were less accurate for data acquired in a different growing season (DBM: RMSEP = 8.50 g plant−1, R2 = 0.77; N-uptake: RMSEP = 0.27 gN plant−1, R2 = 0.68). Recalibration might solve this issue, but additional research is required to cope with this effect during image acquisition and processing. Further improvement of the model robustness could be obtained through the inclusion of phenological parameters such as crop height.

Automatic detection of foliar diseases in potato fields, such as early blight caused by Alternaria solani, could allow farmers to reduce the application of plant protection products while minimizing production losses. UAV-based, high resolution, NIR-sensitive cameras offer the advantage of a detailed top-down perspective, with high-contrast images ideally suited for detecting Alternaria solani lesions. A field experiment was conducted with 8 plots housing 256 infected plants which were monitored 6 times over a 16-day period with a UAV. A modified RGB camera, sensitive to NIR, was combined with a superzoom lens to obtain ultra-high-resolution images with a spatial resolution of 0.3 mm/px. More than 15,000 lesions were annotated with points in two full size images corresponding to 1250 cropped tiles of 256 by 256 pixels. A deep learning U-Net model was trained to predict the density of Alternaria solani lesions for every pixel. In this way, density maps were calculated to indicate disease hotspots as a guide for the farmer.

The objective of this research was to compare the spectral signature of two soil types, Cambisol and Vertisol, cultivated with sugarcane in the Villa Clara province, Cuba. The soil samples were collected at two levels: 1) Landscape level of the province (DC), 2) within a field (CE). The vis-NIR spectral signatures were acquired in the laboratory by means of a diode array spectrophotometer covering the wavelength range from 399 to 1697 nm. The results showed that the absorbance spectra of these two soil types can clearly be distinguished, which suggests that the soil type could be derived from the spectra. The light absorbance by the soil varied with wavelength in each soil type in DC and CE. A visual comparison of the spectral shape and the variation between a field and landscape within the same soil group showed some similitude. Around the wavelength of 1400 nm in both soil types the absorbance values increase and consequently decrease again. These characteristics bring a good match of the spectral signatures of DC and CE. In addition, in CE an increasing pattern of soil absorbance was observed as organic matter (OM %) increases. This pattern was observed around 399 nm and 580 nm while for Cambisol and Vertisol respectively.

The discrimination power of a hyperspectral imaging system for image segmentation or object detection is determined by the illumination, the camera spatial–spectral resolution, and both the pre-processing and analysis methods used for image processing. In this study, we methodically reviewed the alternatives for each of those factors for a case study from the food industry to provide guidance in the construction and configuration of hyperspectral imaging systems in the visible near infrared range for food quality inspection. We investigated both halogen- and LED-based illuminations and considered cameras with different spatial–spectral resolution trade-offs. At the level of the data analysis, we evaluated the impact of binning, median filtering and bilateral filtering as pre- or post-processing and compared pixel-based classifiers with convolutional neural networks for a challenging application in the food industry, namely ingredient identification in a flour–seed mix. Starting from a basic configuration and by modifying the combination of system aspects we were able to increase the mean accuracy by at least 25 %. In addition, different trade-offs in performance-complexity were identified for different combinations of system parameters, allowing adaptation to diverse application requirements.

Today, measurement of raw milk quality and composition relies on Fourier transform infrared spectroscopy to monitor and improve dairy production and cow health. However, these laboratory analyzers are bulky, expensive and can only be used by experts. Moreover, the sample logistics and data transfer delay the information on product quality, and the measures taken to optimize the care and feeding of the cattle render them less suitable for real-time monitoring. An on-farm spectrometer with compact size and affordable cost could bring a solution for this discrepancy. This paper evaluates the performance of microelectromechanical system (MEMS)-based near-infrared (NIR) spectrometers as on-farm milk analyzers. These spectrometers use Fabry–Pérot interferometers for wavelength tuning, giving them the advantage of very compact size and affordable price. This study discusses the ability of MEMS spectrometers to reach the accuracy limits set by the International Committee for Animal Recording (ICAR) for at-line analyzers of the milk content regarding fat, protein and lactose. According to the achieved results, the transmission measurements with the NIRONE 2.5 spectrometer perform best, with an acceptable root mean squared error of prediction (RMSEP = 0.21% w/w) for the measurement of milk fat and excellent performance (RMSEP ≤ 0.11% w/w) for protein and lactose. In addition, the transmission measurements using the NIRONE 2.0 module give similar results for fat and lactose (RMSEP of 0.21 and 0.10% w/w respectively), while the prediction of protein is slightly deteriorated (RMSEP = 0.15% w/w). These results show that the MEMS spectrometers can reach sufficient prediction accuracy compared to ICAR standard values for at-line and in-line fat, protein and lactose prediction.

Recycling of textile materials is becoming important due to the increasing amount of textile waste and its large environmental impact. The Resyntex project aims at dealing with this textile waste by enabling its chemical recycling. To do so, pure textile materials and blends need to be sorted first. In this paper we evaluate the suitability of hyperspectral imaging for pure and blend textile sorting. We also test the discrimination capacity between denim and non-denim textile, since this is required prior to the de-colouration processes. For this purpose, we use a line-scan sensor in the 450–950 nm range, since its cost, compactness and speed characteristics make it suitable for industrial deployment. To deal with the strong colour interference of the textile a hierarchical classification approach is proposed. The results on the available sample set show promising discrimination potential for material discrimination as well as for denim versus non-denim detection.

As plants would benefit from adjusting and optimizing their architecture to changing environmental stimuli, ensuring a strong and healthy plant, it was hypothesized that different soil moisture levels would affect xylem and collenchyma development in basil (Ocimum basilicum L. cv. Marian) stems. Four different irrigation set-points (20, 30, 40 and 50% VWC), corresponding respectively to pF values of 1.95, 1.65, 1.30 and 1.15, were applied. Basil plants grown near the theoretical wilting point (pF 2) had a higher xylem vessel frequency and lower mean vessel diameter, promoting water transport under drought conditions. Cultivation at low soil moisture also impacted the formation of collenchyma in the apical stem segments, providing mechanical and structural support to these fast-growing stems and vascular tissues. The proportion of collenchyma area was significantly lower for the pF1.15 treatment (9.25 ± 3.24%) compared to the pF1.95 and pF1.30 treatments (16.04 ± 1.83% and 13.28 ± 1.38%, respectively). Higher fractions of collenchyma resulted in a higher mechanical stem strength against bending. Additionally, tracheids acted as the major support tissues in the basal stem segments. These results confirm that the available soil moisture impacts mechanical stem strength and overall plant quality of basil plants by impacting xylem and collenchyma development during cultivation, ensuring sufficient mechanical support to the fast-growing stem and to the protection of the vascular tissues. To our knowledge, this study is the first to compare the mechanical and anatomical characteristics of plant stems cultivated at different soil moisture levels.

Stomata, the microscopic pores surrounded by a pair of guard cells on the surfaces of leaves and stems, play an essential role in regulating the gas exchange between a plant and the surrounding atmosphere. Stomatal development and opening are significantly influenced by environmental conditions, both in the short and long term. The rapid rate of current climate change has been affecting stomatal responses, as a new balance between photosynthesis and water-use efficiency has to be found. Understanding the mechanisms involved in stomatal regulation and adjustment provides us with new insights into the ability of stomata to process information and evolve over time. In this review, we summarize the recent advances in research on the underlying mechanisms of the interaction between environmental factors and stomatal development and opening. Specific emphasis is placed on the environmental factors including light, CO2 concentration, ambient temperature, and relative humidity, as these factors play a significant role in understanding the impact of global climate change on plant development.

Close remote sensing approaches can be used for high throughput on-field phenotyping in the context of plant breeding and biological research. Data on canopy cover (CC) and canopy height (CH) and their temporal changes throughout the growing season can yield information about crop growth and performance. In the present study, sigmoid models were fitted to multi-temporal CC and CH data obtained using RGB imagery captured with a drone for a broad set of soybean genotypes. The Gompertz and Beta functions were used to fit CC and CH data, respectively. Overall, 90.4% fits for CC and 99.4% fits for CH reached an adjusted R2 > 0.70, demonstrating good performance of the models chosen. Using these growth curves, parameters including maximum absolute growth rate, early vigor, maximum height, and senescence were calculated for a collection of soybean genotypes. This information was also used to estimate seed yield and maturity (R8 stage) (adjusted R2 = 0.51 and 0.82). Combinations of parameter values were tested to identify genotypes with interesting traits. An integrative approach of fitting a curve to a multi-temporal dataset resulted in biologically interpretable parameters that were informative for relevant traits.

Nowadays, quality inspection of fruit and vegetables is typically accomplished through visual inspection. Automation of this inspection is desirable to make it more objective. For this, hyperspectral imaging has been identified as a promising technique. When the field of view includes multiple objects, hypercubes should be segmented to assign individual pixels to different objects. Unsupervised and supervised methods have been proposed. While the latter are labour intensive as they require masking of the training images, the former are too computationally intensive for in-line use and may provide different results for different hypercubes. Therefore, a semi-supervised method is proposed to train a computationally efficient segmentation algorithm with minimal human interaction. As a first step, an unsupervised classification model is used to cluster spectra in similar groups. In the second step, a pixel selection algorithm applied to the output of the unsupervised classification is used to build a supervised model which is fast enough for in-line use. To evaluate this approach, it is applied to hypercubes of vine tomatoes and table grapes. After first derivative spectral preprocessing to remove intensity variation due to curvature and gloss effects, the unsupervised models segmented 86.11% of the vine tomato images correctly. Considering overall accuracy, sensitivity, specificity and time needed to segment one hypercube, partial least squares discriminant analysis (PLS-DA) was found to be the best choice for in-line use, when using one training image. By adding a second image, the segmentation results improved considerably, yielding an overall accuracy of 96.95% for segmentation of vine tomatoes and 98.52% for segmentation of table grapes, demonstrating the added value of the learning phase in the algorithm.

Despite the research on opportunities to automatically measure lameness in cattle, lameness detection systems are not widely available commercially and are only used on a few dairy farms. However, farmers need to be aware of the lame cows in their herds in order treat them properly and in a timely fashion. Many papers have focused on the automated measurement of gait or behavioral cow characteristics related to lameness. In order for such automated measurements to be used in a detection system, algorithms to distinguish between non-lame and mildly or severely lame cows need to be developed and validated. Few studies have reached this latter stage of the development process. Also, comparison between the different approaches is impeded by the wide range of practical settings used to measure the gait or behavioral characteristic (e.g., measurements during normal farming routine or during experiments; cows guided or walking at their own speed) and by the different definitions of lame cows. In the majority of the publications, mildly lame cows are included in the non-lame cow group, which limits the possibility of also detecting early lameness cases. In this review, studies that used sensor technology to measure changes in gait or behavior of cows related to lameness are discussed together with practical considerations when conducting lameness research. In addition, other prerequisites for any lameness detection system on farms (e.g., need for early detection, real-time measurements) are discussed.

Due to its detrimental effect on cow welfare, health and production, lameness in dairy cows has received quite a lot of attention in the last few decades—not only in terms of prevention and treatment of lameness but also in terms of detection, as early treatment might decrease the number of severely lame cows in the herds as well as decrease the direct and indirect costs associated with lameness cases. Generally, lame cows are detected by the herdsman, hoof trimmer or veterinarian based on abnormal locomotion, abnormal behavior or the presence of hoof lesions during routine trimming. In the scientific literature, several guidelines are proposed to detect lame cows based on visual interpretation of the locomotion of individual cows (i.e., locomotion scoring systems). Researchers and the industry have focused on automating such observations to support the farmer in finding the lame cows in their herds, but until now, such automated systems have rarely been used in commercial herds. This review starts with the description of normal locomotion of cows in order to define ‘abnormal’ locomotion caused by lameness. Cow locomotion (gait and posture) and behavioral features that change when a cow becomes lame are described and linked to the existing visual scoring systems. In addition, the lack of information of normal cow gait and a clear description of ‘abnormal’ gait are discussed. Finally, the different set-ups used during locomotion scoring and their influence on the resulting locomotion scores are evaluated.

Most automatic lameness detection system prototypes have not yet been commercialized, and are hence not yet adopted in practice. Therefore, the objective of this study was to simulate the effect of detection performance (percentage missed lame cows and percentage false alarms) and system cost on the potential market share of three automatic lameness detection systems relative to visual detection: a system attached to the cow, a walkover system, and a camera system. Simulations were done using a utility model derived from survey responses obtained from dairy farmers in Flanders, Belgium. Overall, systems attached to the cow had the largest market potential, but were still not competitive with visual detection. Increasing the detection performance or lowering the system cost led to higher market shares for automatic systems at the expense of visual detection. The willingness to pay for extra performance was €2.57 per % less missed lame cows, €1.65 per % less false alerts, and €12.7 for lame leg indication, respectively. The presented results could be exploited by system designers to determine the effect of adjustments to the technology on a system’s potential adoption rate.

Packaging protects food products from environmental influences, assuring quality and safety throughout shelf life if properly performed. Packaging quality depends on the quality of the packaging material and of the closure or seal. A common problem possibly jeopardizing seal quality is the presence of seal contamination, which can cause a decreased seal strength, an increased packaging failure risk and leak formation. Therefore, early detection and removal of seal contaminated packages from the production chain is crucial. In this work, a pulsed-type active thermography method using the heated seal bars as an excitation source was studied for detecting seal contamination. Thermal image sequences of contaminated seals were recorded shortly after sealing. The detection performances of six thermal image processing methods, based on a single frame, a fit of the cooling profiles, thermal signal reconstruction, pulsed phase thermography, principal component thermography and a matched filter, were compared. High resolution digital images served as a reference to quantify seal contamination, and processed thermal images were mapped to these references. The lowest detection limit (equivalent diameter 0.60 mm) was obtained for the method based on a fit of the cooling profiles. Moreover, the detection performance of this method did not depend strongly on the time after sealing at which recording of the thermal images was started, making it a robust and generally applicable method.