Leaning on Smart Agricultural Systems for Crop Monitoring

Contemporary precision agriculture and breeding programs are heavily dependent on the evaluation of a high number of experimental plots, usually involving different genotypes, irrigation methods, and managing systems [1]. Crop improvement programs vary widely in size and structure, usually depending on their research scope [2,3]. Public-sector programs tend to have broader objectives, often including food security and nutritional improvement, while private-sector breeding is often driven by commercial interests and profitability [4]. In both cases, they may target multiple crop species and operate across diverse locations and environments. In this context, phenotyping approaches need to be accurate and precise in order to produce comparable measures within the experimental field and between locations while maintaining a high-throughput capacity [5]. However, the existing conventional approaches of crop monitoring often rely on visual evaluation and manual data collection, making them labor-intensive, time-consuming, and resource-demanding, and thus, the process presents significant limitations that hinder the scalability and efficiency of these efforts [6]. Even when non-destructive methods are used, subjective visual assessments introduce variability due to observer bias and inconsistencies between evaluators. The bottleneck created by slow, inconsistent, or incomplete data collection can delay the breeding cycle and reduce the effectiveness of selection decisions [7,8].

Accurate monitoring of these field trials requires significant human and technical resources, often involving large teams of trained personnel. This demand for resources can limit both the scale and precision of field-based activities [9]. In this context, there is a growing need for innovative and objective crop monitoring solutions. Emerging smart agriculture systems, integrating advanced sensing technologies, automation, and data analytics, offer a promising approach to address these challenges [10]. These technologies support real-time, high-resolution, and scalable crop evaluation, helping to enhance decision-making and increase the efficiency of crop evaluation [11].

Smart agriculture relies heavily on remote sensing, which plays an important role by offering tools to increase throughput, accuracy, and efficiency in large-scale experimental studies, consequently providing information about crop health, soil conditions, and environmental factors, which is crucial in precision farming and sustainable agricultural practices. For instance, satellite crop monitoring has been largely reported as a powerful technology that enables real-time tracking of crop development through spectral analysis [12] and highlights the causes of spatial variability [13]. Key developments include the integration of unmanned aerial vehicles (UAVs) equipped with multispectral, hyperspectral, or thermal cameras, which can capture spatial and temporal variability in crop performance at fine resolution and across entire fields within minutes. These platforms enable non-destructive and repeatable monitoring of traits such as canopy development, senescence, biomass accumulation, and stress responses. In addition, ground-based platforms play a key role in capturing detailed crop physiological responses, providing high-resolution data at the plant or plot level [14]. Meanwhile, sensor networks integrated through Internet of Things (IoT) technologies enable continuous monitoring of environmental parameters, including soil moisture, temperature, and light interception [15]. This holistic monitoring approach is crucial for understanding crop–environment interactions, optimizing resource use, and accelerating the selection of resilient, high-performing genotypes in field experimentation.

Smart crop monitoring devices provide real-time data and actionable insights that support informed decision-making, leading to more efficient input use and improved sustainability in agriculture [16]. For example, nutrient management has become a key focus in smart farming, shifting from generalized fertilizer recommendations to precision systems driven by real-time data, remote sensing, and decision-support algorithms [17]. This transformation is particularly impactful in nitrogen management, where in-season diagnostic models enable targeted, need-based fertilizer applications [18]. When integrated with farmer-friendly decision-support tools and open access platforms, these innovations help translate experimental research into practical solutions. By combining satellite data with agronomic models, they produce spatially variable nitrogen prescription maps customized for individual farms and ready for immediate use [19,20]. Beyond optimizing fertilization and anticipating nutrient deficiencies, these systems also enable early detection of crop diseases [21] and water stress [22], supporting automated decisions on irrigation and pesticide use. Machine learning has further advanced precision agriculture by integrating complex datasets (such as soil properties, weather patterns, remote sensing outputs, and historical management practices) to enable site-specific and optimized crop management through a multidisciplinary approach [23]. This shift from uniform to spatially and temporally adaptive management enhances the resilience and efficiency of cropping systems. It supports sustainable intensification, highlighting the transformative impact of smart resource management in modern agronomy.

However, the question remains: can these technologies fully replace in-field manual and visual evaluations? A realistic perspective on smart agriculture systems should focus on their role as complementary tools and approaches to lean on to reduce costs in time and resources, but not as complete substitutes [9,24]. For instance, the high initial costs associated with the technology and data processing, together with the need for digital skill, are existing barriers that might still be limiting the full potential of existing smart agriculture technologies among farmers [25,26]. When combined with artificial intelligence (AI) and machine learning algorithms, these systems can process large volumes of data to extract meaningful phenotypic traits, identify patterns, and generate predictive models with minimal human intervention [27]. One challenge in using artificial intelligence models for crop monitoring is the variability between years due to weather and management practices [28,29]. Even though large datasets collected over multiple locations and years might seem advantageous for training machine learning algorithms, they can reduce a model’s ability to generalize accurately to current conditions. For instance, a model trained on data from a rainier year may misinterpret crop development patterns during a dry, hot year. These shifts introduce noise and variability into the model, reducing its reliability when applied to a new season with different conditions. Although recent years have seen efforts to introduce transfer learning approaches for prediction across environments and years [30], a practical solution could be introducing data collected within the same growing season to the trained model in order to encourage better alignment with the environment, management strategies, and current crop behavior [31].

Together, these innovations are shaping the future of precision phenotyping by enabling breeding programs and agronomic research to scale their operations, reduce human bias and labor requirements, and enhance the timeliness and objectivity of trait evaluation [32,33]. The Special Issue titled “Smart Agriculture Sensors and Monitoring Systems for Field Detection” showcases impactful examples of how emerging technologies represent a transformative shift toward data-driven decision-making in crop improvement and field experimentation. It includes case studies involving the monitoring of crops at different scales, including an assessment of blast incidence in rice using the Sentinel-2 satellite [34], an evaluation of water-use efficiency in cotton [35] and durum wheat [36] using hand-held tools, the use of advanced imaging techniques to analyze morphometric fruit characteristics [37], rice architecture [38], and lettuce height [39], and the placement of in-field sensors to detect avian pests [40].