Real-Time AI-Assisted Push-Broom Hyperspectral System for Precision Agriculture

In the ever-evolving landscape of modern agriculture, the integration of advanced technologies has become indispensable for optimizing crop management and ensuring sustainable food production. This paper presents the development and implementation of a real-time AI-assisted push-broom hyperspectral system for plant identification. The push-broom hyperspectral technique, coupled with artificial intelligence, offers unprecedented detail and accuracy in crop monitoring. This paper details the design and construction of the spectrometer, including optical assembly and system integration. The real-time acquisition and classification system, utilizing an embedded computing solution, is also described. The calibration and resolution analysis demonstrates the accuracy of the system in capturing spectral data. As a test, the system was applied to the classification of plant leaves. The AI algorithm based on neural networks allows for the continuous analysis of hyperspectral data relative up to 720 ground positions at 50 fps.

Automated Prototype for Bombyx mori Cocoon Sorting Attempts to Improve Silk Quality and Production Efficiency through Multi-Step Approach and Machine Learning Algorithms

Cocoon sorting is one of the most labor-demanding activities required both at the end of the agricultural production and before the industrial reeling process to obtain an excellent silk quality. In view of the possible relaunch of European sericulture, the automatization of this production step is mandatory both to reduce silk costs and to standardize fiber quality. The described research starts from this criticality in silk production (the manual labor required to divide cocoons into different quality classes) to identify amelioration solutions. To this aim, the automation of this activity was proposed, and a first prototype was designed and built. This machinery is based on the use of three cameras and imaging algorithms identifying the shape and size of the cocoons and outside stains, a custom-made light sensor and an AI model to discard dead cocoons. The current efficiency of the machine is about 80 cocoons per minute. In general, the amelioration obtained through this research involves both the application of traditional sensors/techniques to an unusual product and the design of a dedicated sensor for the identification of dead/alive pupae inside the silk cocoons. A general picture of the overall efficiency of the new cocoon-sorting prototype is also outlined.

Olive Fruit Selection through AI Algorithms and RGB Imaging

(1) Background: Extra virgin olive oil production is strictly influenced by the quality of fruits. The optical selection allows for obtaining high quality oils starting from batches with different qualitative characteristics. This study aims to test a CNN algorithm in order to assess its potential for olive classification into several quality classes for industrial purposes, specifically its potential integration and sorting performance evaluation. (2) Methods: The acquired samples were all subjected to visual analysis by a trained operator for the distinction of the products in five classes related to the state of external veraison and the presence of visible defects. The olive samples were placed at a regular distance and in a fixed position on a conveyor belt that moved at a constant speed of 1 cm/s. The images of the olives were taken every 15 s with a compact industrial RGB camera mounted on the main frame in aluminum to allow overlapping of the images, and to avoid loss of information. (3) Results: The modelling approaches used, all based on AI techniques, showed excellent results for both RGB datasets. (4) Conclusions: The presented approach regarding the qualitative discrimination of olive fruits shows its potential for both sorting machine performance evaluation and for future implementation on machines used for industrial sorting processes.

Superior EVOO Quality Production: An RGB Sorting Machine for Olive Classification

Extra virgin olive oil (EVOO) is a commercial product of high quality, thanks to its nutritional and organoleptic characteristics. The olives ripeness and the choice of harvest time according to their color and size, strongly influences the quality of the EVOO. The physical sorting of olives with machines performing rapid and objective optical selection, impossible by hand, can improve the quality of the final product. The aim of this study concerns the classification of olives into two qualitative classes, based on the maturity stage and the presence of external defects, through an industrial RGB optical sorting prototype, evaluating its performance and comparing the results with those obtained visually by trained operators. EVOOs obtained from classified olives were characterized through chemical, physical-chemical analysis and sensory profile. For the first time, the optoelectronic technologies in an industrial system was tested on olives to produce superior quality EVOO. The selection allows late harvest, obtaining oils with good characteristics from fully ripe and unripe fruits together, separating defective olives with appropriate calibration and training. Optoelectronic selection creates the opportunity to blend the obtained oils destined to different applications according to the needs of the consumer or producer, using a vanguard technology at low cost.

Advantages in Using Colour Calibration for Orthophoto Reconstruction

UAVs are sensor platforms increasingly used in precision agriculture, especially for crop and environmental monitoring using photogrammetry. In this work, light drone flights were performed on three consecutive days (with different weather conditions) on an experimental agricultural field to evaluate the photogrammetric performances due to colour calibration. Thirty random reconstructions from the three days and six different areas of the field were performed. The results showed that calibrated orthophotos appeared greener and brighter than the uncalibrated ones, better representing the actual colours of the scene. Parameter reporting errors were always lower in the calibrated reconstructions and the other quantitative parameters were always lower in the non-calibrated ones, in particular, significant differences were observed in the percentage of camera stations on the total number of images and the reprojection error. The results obtained showed that it is possible to obtain better orthophotos, by means of a calibration algorithm, to rectify the atmospheric conditions that affect the image obtained. This proposed colour calibration protocol could be useful when integrated into robotic platforms and sensors for the exploration and monitoring of different environments.

Sorting biotic and abiotic stresses on wild rocket by leaf-image hyperspectral data mining with an artificial intelligence model

BACKGROUND

Wild rocket (Diplotaxis tenuifolia) is prone to soil-borne stresses under intensive cultivation systems devoted to ready-to-eat salad chain, increasing needs for external inputs. Early detection of the abiotic and biotic stresses by using digital reflectance-based probes may allow optimization and enhance performances of the mitigation strategies.

METHODS

Hyperspectral image analysis was applied to D. tenuifolia potted plants subjected, in a greenhouse experiment, to five treatments for one week: a control treatment watered to 100% water holding capacity, two biotic stresses: Fusarium wilting and Rhizoctonia rotting, and two abiotic stresses: water deficit and salinity. Leaf hyperspectral fingerprints were submitted to an artificial intelligence pipeline for training and validating image-based classification models able to work in the stress range. Spectral investigation was corroborated by pertaining physiological parameters.

RESULTS

Water status was mainly affected by water deficit treatment, followed by fungal diseases, while salinity did not change water relations of wild rocket plants compared to control treatment. Biotic stresses triggered discoloration in plants just in a week after application of the treatments, as evidenced by the colour space coordinates and pigment contents values. Some vegetation indices, calculated on the bases of the reflectance data, targeted on plant vitality and chlorophyll content, healthiness, and carotenoid content, agreed with the patterns of variations observed for the physiological parameters. Artificial neural network helped selection of VIS (492-504, 540-568 and 712-720 nm) and NIR (855, 900-908 and 970 nm) bands, whose read reflectance contributed to discriminate stresses by imaging.

CONCLUSIONS

This study provided significative spectral information linked to the assessed stresses, allowing the identification of narrowed spectral regions and single wavelengths due to changes in photosynthetically active pigments and in water status revealing the etiological cause.

Current Methods, Common Practices, and Perspectives in Tracking and Monitoring Bioinoculants in Soil

Microorganisms promised to lead the bio-based revolution for a more sustainable agriculture. Beneficial microorganisms could be a valid alternative to the use of chemical fertilizers or pesticides. However, the increasing use of microbial inoculants is also raising several questions about their efficacy and their effects on the autochthonous soil microorganisms. There are two major issues on the application of bioinoculants to soil: (i) their detection in soil, and the analysis of their persistence and fate; (ii) the monitoring of the impact of the introduced bioinoculant on native soil microbial communities. This review explores the strategies and methods that can be applied to the detection of microbial inoculants and to soil monitoring. The discussion includes a comprehensive critical assessment of the available tools, based on morpho-phenological, molecular, and microscopic analyses. The prospects for future development of protocols for regulatory or commercial purposes are also discussed, underlining the need for a multi-method (polyphasic) approach to ensure the necessary level of discrimination required to track and monitor bioinoculants in soil.

An Open Source Low-Cost Device Coupled with an Adaptative Time-Lag Time-Series Linear Forecasting Modeling for Apple Trentino (Italy) Precision Irrigation

Precision irrigation represents those strategies aiming to feed the plant needs following the soil’s spatial and temporal characteristics. Such a differential irrigation requires a different approach and equipment with regard to conventional irrigation to reduce the environmental impact and the resources use while maximizing the production and thus profitability. This study described the development of an open source soil moisture LoRa (long-range) device and analysis of the data collected and updated directly in the field (i.e., weather station and ground sensor). The work produced adaptive supervised predictive models to optimize the management of agricultural precision irrigation practices and for an effective calibration of other agronomic interventions. These approaches are defined as adaptive because they self-learn with the acquisition of new data, updating the on-the-go model over time. The location chosen for the experimental setup is a cultivated area in the municipality of Tenna (Trentino, Alto Adige region, Italy), and the experiment was conducted on two different apple varieties during summer 2019. The adaptative partial least squares time-lag time-series modeling, in operative field conditions, was a posteriori applied in the consortium for 78 days during the dry season, producing total savings of 255 mm of irrigated water and 44,000 kW of electricity, equal to 10.82%.

Light Drones for Basic In-Field Phenotyping and Precision Farming Applications: RGB Tools Based on Image Analysis

Plant phenotyping has garnered major attention in recent years, leading to developing new strategies to measure and assess plant traits of interest. For data acquisition of large fields, devices and sensors are required that deliver detailed and reproducible temporal and spatial information on the cultivated crop. This work proposes the potential use of low-cost light drones for in-field phenotyping applications on cereal crops. The proposed method allows to obtain precise measurements of color and height of the plants for the individual plots. The method is based on a color calibration algorithm (TPS-3D interpolating function) and a 3D ortho image reconstruction. The method has been applied on an experimental field with durum and soft wheat parcels obtaining information on real color (with an error lower than 12/256) and height for each single plot.

Internet of Beer: A Review on Smart Technologies from Mash to Pint

The beer production chain includes some crucial steps regarding processing, delivery, service, and consumption that can benefit from the implementation of IoT (Internet of Things) based technologies. Large breweries implemented the use of sensors and digitization before smaller ones among which are craft breweries. Internet of Beer (IoB) technologies are becoming accessible to mid and small sized brewing companies. Therefore, the objective of this work is to review mainly low-cost IoB smart technologies that can be implemented from the mash to the final product and its service, to improve the brewing production, control, delivery, and final quality increasing profitability. The reviewed applications were retrieved both from the scientific databases and from the web. The work is structured in three macro areas such as beer processing, product logistics and traceability, and service. The results show a future trend characterized by a very fast increase in the use of IoB (also open source) systems to drive efficiency, productivity, quality, and safety. This will be done by real-time monitoring and a data-driven decision support system (DSS). Crucial aspects needing further investigation are data ownership and data standardization. The access price of IoB devices and software is destined for a significant decrease while their diversification on the market will grow leading to a massive future implementation within all the production levels.

A Full Technological Traceability System for Extra Virgin Olive Oil

The traceability of extra virgin olive oil (EVOO) could guarantee the authenticity of the product and the protection of the consumer if it is part of a system able to certify the traceability information. The purpose of this paper was to propose and apply a complete electronic traceability prototype along the entire EVOO production chain of a small Italian farm and to verify its economic sustainability. The full traceability of the EVOO extracted from 33 olive trees from three different cultivars (Carboncella, Frantoio and Leccino) was considered. The technological traceability system (TTS; infotracing) consists of several open source devices (based on radio frequency identification (RFID) and QR code technologies) able to track the EVOO from the standing olive tree to the final consumer. The infotracing system was composed of a dedicated open source app and was designed for easy blockchain integration. In addition, an economic analysis of the proposed TTS, with reference to the semi-mechanized olive harvesting process, was conducted. The results showed that the incidence of the TTS application on the whole production varies between 3% and 15.5%, (production from 5 to 60 kg tree⁻¹). The application at the consortium level with mechanized harvesting is fully sustainable in economic terms. The proposed TTS could not only provide guarantees to the final consumer but could also direct the farmer towards precision farming management.

Light Drone-Based Application to Assess Soil Tillage Quality Parameters

The evaluation of soil tillage quality parameters, such as cloddiness and surface roughness produced by tillage tools, is based on traditional methods ranging, respectively, from manual or mechanical sieving of ground samples to handheld rulers, non-contact devices or Precision Agriculture technics, such as laser profile meters. The aim of the study was to compare traditional methods of soil roughness and cloddiness assessment (laser profile meter and manual sieving), with light drone RGB 3D imaging techniques for the evaluation of different tillage methods (ploughed, harrowed and grassed). Light drone application was able to replicate the results obtained by the traditional methods, introducing advantages in terms of time, repeatability and analysed surface while reducing the human error during the data collection on the one hand and allowing a labour-intensive field monitoring solution for digital farming on the other. Indeed, the profilometer positioning introduces errors and may lead to false reading due to limited data collection. Future work could be done in order to streamline the data processing operation and so to produce a practical application ready to use and stimulate the adoption of new evaluation indices of soil cloddiness, such as Entropy and the Angular Second Moment (ASM), which seem more suitable than the classic ones to achieved data referred to more extended surfaces.

A review on blockchain applications in the agri-food sector

BACKGROUND

Food security can benefit from the technology's transparency, relatively low transaction costs and instantaneous applications. A blockchain is a distributed database of records in the form of encrypted blocks, or a public ledger of all transactions or digital events that have been executed and shared among participating parties and can be verified at any time in the future. Generally, the robust and decentralized functionality of the blockchain is used for global financial systems, but it can easily be expanded to contracts and operations such as tracking of the global supply chain. In the precision agriculture context, Information and Communications Technology can be further implemented with a blockchain infrastructure to enable new farm systems and e-agriculture schemes.

RESULTS

The purpose of this review is to show a panorama of the scientific studies (enriched by a terms mapping analysis) on the use of blockchain in the agri-food sector, from both an entirely computational and an applicative point of view. As evidenced by the network analysis, the reviewed studies mainly focused on software aspects (e.g. the architecture and smart contracts). However, some aspects regarding the different blockchain knots (computers always connected to the blockchain network) having the role to store and distribute an updated copy of each block in a food supply-chain, result of crucial importance.

CONCLUSION

These technologies appear very promising and rich of great potential showing a good flexibility for applications in several sectors but still immature and hard to apply due to their complexity. © 2019 Society of Chemical Industry.

Are the Innovative Electronic Labels for Extra Virgin Olive Oil Sustainable, Traceable, and Accepted by Consumers?

Traceability is the ability to follow the displacement of food through its entire chain. Extra virgin olive oil (EVOO) represents Italian excellence, with consumers’ increased awareness for traceability. The aim of this work is to propose and analyze the economic sustainability and consumers’ preference of three technological systems supporting traceability: Near Field Communication (NFC) based; tamper-proof device plus Radio Frequency Identification (RFID) and app; QR code tag plus “scratch and win” system and blockchain. An anonymous questionnaire to Italian consumers (n = 1120) was made to acquire consumers’ acceptability of the systems and estimating their willingness to pay additional premium prices for these. An economic analysis estimated and compared the technology costs at different production levels. Results show that 94% of the consumer respondents are interested in the implementation of such technologies, and among them 45% chose QR-code protected by a “scratch-and-win” system with a blockchain infotracing-platform (QR-B). The consumers interested are willing to pay a mean premium price of 17.8% and economic analysis reported evidenced an incidence always lower than mid-/high-production levels. The success of the QR-B could be ascribed to different aspects: the cutting-edge fashion trend of blockchain in the food sector, the use of incentives, the easy-to-use QR-code, and the gamification strategy.

Rapid assessment of As and other elements in naturally-contaminated calcareous soil through hyperspectral VIS-NIR analysis

Although arsenic (As) toxicity in soil vary depending on its chemical forms and oxidation states, regulatory limits for this compartment rely on total As content. Conventional methods of total As determination are expensive and time-consuming. The development of predictive techniques might enable a speditive assessment of As contamination in those scenarios, such as thermal spring sites, where exposure to the metalloid poses a threat to human health. The objective of this study was to assess the suitability of Visible Near Infrared spectrophotometry for predicting the total As content in highly calcareous thermal spring soils and the same aim was pursued for those elements (i.e. Al, Fe and Mn) the chemistry of which is tightly connected with that of As. A Partial Least Square approach, including cross-validation and external independent test, was used to relate the concentrations of the target elements to spectral data. The most accurate prediction was found for As with Pearson's coefficient, RMSE, RPD and SEP being equal to 0.94, 69.65, 2.9 and 66.99, respectively. Less accurate predictions were found for Al (r = 0.88; RMSE = 11014; RPD = 1.96; SEP = 11014), Fe (r = 0.93; RMSE = 6921.1; RPD = 2.45; SEP = 6462.4), and Mn (r = 0.92; RMSE = 542.01; RPD = 2.43; SEP = 529.79).

Assessment of quality-assured Tarocco orange fruit sorting rules by combined physicochemical and sensory testing

BACKGROUND

The aim of this study was to extract a sorting rule for Tarocco orange fruit from several physicochemical and sensory tests performed on a marketable lot of 399 Tarocco orange fruits.

RESULTS

The elastic tension at 5% strain (T₅ ) was found to be linearly correlated (r = 0.65) with the Magness-Taylor (MT) index. Thus T₅ was regarded as a non-destructive parameter quantifying fruit firmness and used to categorise the aforementioned lot in three different firmness classes, high (HF), medium (MF) and low (LF). Only the MT index, fruit rind thickness near the fruit peduncle, lightness coefficient and yellow/blue hue component of the orange flesh, as well as total soluble solid content, confirmed the validity of this discrimination at the significance level of 5%. Sensory professionals recognised the greater compactness (7 ± 2) but lower ease of peeling (4 ± 2) and segment separation (4 ± 2) of the HF oranges with respect to the corresponding sensory attributes of orange fruits grouped in the MF and LF classes.

CONCLUSION

To limit the costly rejection of Tarocco orange fruit considered too soft, especially after long-term shipping, it would be reasonable to select only fruits characterised by a compressive force or tension at 5% strain in the range 23-41 N or 300-540 N m⁻¹ respectively.

Electronic nose application for determination of Penicillium digitatum in Valencia oranges

BACKGROUND

Penicillium digitatum and Penicillium italicum are responsible for one the most serious diseases occurring during storage of citrus fruits. Its early detection allows a relevant increase in shelf life, and in situ monitoring of fungal infections represents a very efficient tool to improve storage quality. In the case of metabolic alterations due to physiological or fungal pathologies, olfactometric analysis allows the detection of specific volatile biomarkers, thus providing an effective tool for postharvest quality control of fruits and vegetables.

RESULTS

A total of 300 Valencia oranges were analysed with an electronic nose and results were screened by a multivariate classification technique, partial least squares discriminant analysis, in order to investigate whether the electronic nose could distinguish between Penicillium-infected and non-infected samples and to evaluate the efficiency of the group classifications. High percentages of correct classification were obtained at low levels of infection (100% for 2-5% infection in an independent test).

CONCLUSION

The electronic nose may be successfully applied as a reliable, non-destructive and non-contact indirect technology for the identification of fungal strains in storage rooms, especially when the infection occurs in small percentages that are not easily identifiable by classic methodologies of inspection.

RGB color calibration for quantitative image analysis: the "3D thin-plate spline" warping approach

In the last years the need to numerically define color by its coordinates in n-dimensional space has increased strongly. Colorimetric calibration is fundamental in food processing and other biological disciplines to quantitatively compare samples' color during workflow with many devices. Several software programmes are available to perform standardized colorimetric procedures, but they are often too imprecise for scientific purposes. In this study, we applied the Thin-Plate Spline interpolation algorithm to calibrate colours in sRGB space (the corresponding Matlab code is reported in the Appendix). This was compared with other two approaches. The first is based on a commercial calibration system (ProfileMaker) and the second on a Partial Least Square analysis. Moreover, to explore device variability and resolution two different cameras were adopted and for each sensor, three consecutive pictures were acquired under four different light conditions. According to our results, the Thin-Plate Spline approach reported a very high efficiency of calibration allowing the possibility to create a revolution in the in-field applicative context of colour quantification not only in food sciences, but also in other biological disciplines. These results are of great importance for scientific color evaluation when lighting conditions are not controlled. Moreover, it allows the use of low cost instruments while still returning scientifically sound quantitative data.

Development of a rapid soil water content detection technique using active infrared thermal methods for in-field applications

The aim of this study was to investigate the suitability of active infrared thermography and thermometry in combination with multivariate statistical partial least squares analysis as rapid soil water content detection techniques both in the laboratory and the field. Such techniques allow fast soil water content measurements helpful in both agricultural and environmental fields. These techniques, based on the theory of heat dissipation, were tested by directly measuring temperature dynamic variation of samples after heating. For the assessment of temperature dynamic variations data were collected during three intervals (3, 6 and 10 s). To account for the presence of specific heats differences between water and soil, the analyses were regulated using slopes to linearly describe their trends. For all analyses, the best model was achieved for a 10 s slope. Three different approaches were considered, two in the laboratory and one in the field. The first laboratory-based one was centred on active infrared thermography, considered measurement of temperature variation as independent variable and reported r = 0.74. The second laboratory–based one was focused on active infrared thermometry, added irradiation as independent variable and reported r = 0.76. The in-field experiment was performed by active infrared thermometry, heating bare soil by solar irradiance after exposure due to primary tillage. Some meteorological parameters were inserted as independent variables in the prediction model, which presented r = 0.61. In order to obtain more general and wide estimations in-field a Partial Least Squares Discriminant Analysis on three classes of percentage of soil water content was performed obtaining a high correct classification in the test (88.89%). The prediction error values were lower in the field with respect to laboratory analyses. Both techniques could be used in conjunction with a Geographic Information System for obtaining detailed information on soil heterogeneity.

Nitrogen concentration estimation in tomato leaves by VIS-NIR non-destructive spectroscopy

Nitrogen concentration in plants is normally determined by expensive and time consuming chemical analyses. As an alternative, chlorophyll meter readings and N-NO₃ concentration determination in petiole sap were proposed, but these assays are not always satisfactory. Spectral reflectance values of tomato leaves obtained by visible-near infrared spectrophotometry are reported to be a powerful tool for the diagnosis of plant nutritional status. The aim of the study was to evaluate the possibility and the accuracy of the estimation of tomato leaf nitrogen concentration performed through a rapid, portable and non-destructive system, in comparison with chemical standard analyses, chlorophyll meter readings and N-NO₃ concentration in petiole sap. Mean reflectance leaf values were compared to each reference chemical value by partial least squares chemometric multivariate methods. The correlation between predicted values from spectral reflectance analysis and the observed chemical values showed in the independent test highly significant correlation coefficient (r = 0.94). The utilization of the proposed system, increasing efficiency, allows better knowledge of nutritional status of tomato plants, with more detailed and sharp information and on wider areas. More detailed information both in space and time is an essential tool to increase and stabilize crop quality levels and to optimize the nutrient use efficiency.