Introduction

The collection and management of plant genetic resources are critical to ensuring sustainable food security and enhancing agricultural productivity in the face of a rapidly growing global population and accelerating climate change (Hoban et al., 2021; Jarvis et al., 2010). These efforts serve as foundational strategies for the continued prosperity of humankind. Effective conservation and utilization of genetic resources provide essential data for the development of climate-resilient crop varieties and the design of precision agriculture systems under dynamic environmental conditions (Li et al., 2014; Pieruschka and Schurr, 2019).

In particular, recent advancements in genome sequencing technologies and the accumulation of genomic data have dramatically increased the availability of genotypic information (Poland and Rife, 2012). However, the parallel accumulation of phenotypic data has lagged behind due to the significant resources required for phenotyping, which remains a major bottleneck in the full utilization of genetic resources (Nguyen et al., 2025; Song et al., 2021). Understanding the complex relationship between genotype and phenotype is not only essential for identifying whether observed traits are genetically determined or environmentally induced, but also crucial for accelerating the selection of new cultivars and optimizing agro-environmental interventions in precision farming (Danilevicz et al., 2022; Egea-Gilabert et al., 2021; Qiao et al., 2022).

In Korea, the Rural Development Administration (RDA) has taken strategic steps to address these challenges by developing the “Seed Identification Card” model for key crops such as soybean and buckwheat, aiming to enhance the accessibility and utilization of national germplasm collections (Kim, 2019; Kim et al., 2021b). In addition, a domestic study applying LiDAR-based point cloud data in sorghum demonstrated high accuracy in quantifying plant height and panicle traits, highlighting the potential of LiDAR as a high-throughput phenotyping (HTP) technology for precise and rapid morphological analysis (Ajay Patel et al., 2022). HTP is also being increasingly explored in Korea as a tool to detect drought-induced phenotypic changes in crops, offering a non-invasive and scalable approach for stress physiology studies (Kim et al., 2021a). UAV-RGB imaging combined with structure-from-motion algorithms has been applied to accurately estimate plant height in kenaf, enabling the identification of high-yielding cultivars through aerial phenotyping (Jang et al., 2022). Together, these cases exemplify Korea’s active adoption of advanced phenotyping technologies to enhance the utilization of national genetic resources in agriculture.

This review explores the advantages of standardized image-based analysis of genetic resources, with a focus on how such approaches have improved the efficiency and precision of biological and breeding research in recent years.

To contextualize these advances, we first examine the limitations of conventional phenotyping methods and the emergence of high-throughput phenotyping (HTP) as a transformative solution.

Limitations of Traditional Phenotyping and the Emergence of High-Throughput Phenotyping with the Need for Standardization

Conventional plant phenotyping has historically relied on manual measurements and visual assessments performed by researchers. While this approach has contributed significantly to the understanding of plant traits, it is inherently time-consuming, labor-intensive, and subject to observer bias (Omari et al., 2020). Furthermore, traditional phenotyping often involves destructive sampling methods, which prevent continuous monitoring of individual plants and hinder dynamic analyses of developmental processes over time (Omari et al., 2020). Table 1 provides a comparative overview of conventional phenotyping and high-throughput phenotyping (HTP), highlighting their methodological, operational, and analytical distinctions.

To overcome these limitations, High-Throughput Phenotyping (HTP) technologies have been introduced (Li et al., 2014). HTP systems automate phenotypic data acquisition using a range of imaging sensors and analytical tools, enabling more accurate, consistent, and large-scale measurements with increased efficiency (Omari et al., 2020). These technologies have opened new avenues in agricultural innovation and serve as a cornerstone of smart farming practices by facilitating precise monitoring of plant performance under diverse environmental conditions.

Despite these advances, the accumulation of massive phenotypic datasets without a unified framework poses new challenges. Disorganized or unstandardized data can become a barrier rather than a resource, complicating biological interpretations and limiting the effective use of genetic resources in breeding programs. Given the sheer volume and diversity of data generated by HTP platforms, establishing standardized data formats and trait ontologies becomes imperative to ensure interoperability, consistency, and cross-study comparability. To fully harness the potential of HTP and maximize the value of genetic resources, it is essential to establish integrative platforms that combine genotypic and phenotypic data, underpinned by standardized protocols and data structures.

One such critical component is the adoption of standardized trait ontologies. Ontologies provide a structured vocabulary that enables efficient data organization, interoperability across databases, and the ability to compare phenotypic data collected by different researchers, in different regions, and across different years (Volk, 2010). In Korea, the “Seed Identification Card” model developed by the Rural Development Administration (RDA) serves as a domestic example of standardizing plant trait descriptors (Kim, 2019; Kim et al., 2021b). Internationally, efforts such as the European Cooperative Programme for Plant Genetic Resources (ECPGR) illustrate how standardized systems can enhance the accessibility and usability of genetic resource pools (Volk, 2010).

Standardized image-based phenotypic data go beyond mere data archiving. They allow for cross-comparison and integrated analysis of heterogeneous datasets, thereby enabling the large-scale investigation of complex biological phenomena such as genotype-by-environment interactions. Without such standardization, comparative and meta-analytical approaches across large, multi-location, and long-term studies become nearly impossible (Volk, 2010).

Moreover, standardized phenotypic datasets serve as a foundation for sophisticated genomic analyses such as Genome-Wide Association Studies (GWAS) and Genomic Selection (GS) (Li et al., 2014). These tools empower breeders to identify, select, and combine desirable traits with greater precision and speed. Ultimately, this accelerates the development of superior cultivars that are better adapted to changing environmental and socio-economic conditions, thereby supporting the broader goals of sustainable agriculture.

Accelerating Biological and Breeding Research through Image-Based Phenotyping

High-throughput phenotyping (HTP) has emerged as a transformative force in modern breeding programs, enabling the efficient evaluation of large and diverse populations while maximizing biological insights for agricultural advancement (Araus and Cairns, 2014; Cabrera-Bosquet et al., 2012). By integrating HTP technologies into breeding pipelines, researchers can now assess more genotypes within limited timeframes and resources, thereby accelerating the identification and selection of superior individuals (Nguyen et al., 2025).

More specifically, the application of image-based phenotyping has proven critical not only for improving yield potential but also for the development of crop varieties with enhanced resistance to biotic and abiotic stresses (Angidi et al., 2025; Singh et al., 2016). HTP facilitates the quantitative measurement of complex traits at physiological, morphological, structural, biochemical, and molecular levels, and generates standardized, high-quality phenotypic datasets necessary for advanced genomic-assisted breeding techniques such as genome-wide association studies (GWAS), linkage mapping, and marker-assisted selection (MAS) (Kumar and Kaushik, 2023; Li et al., 2014). These data address key bottlenecks in breeding by providing robust, reproducible trait measurements, thereby accelerating the release of improved cultivars.

Moreover, HTP reduces dependency on manual labor and subjective observations inherent in traditional phenotyping methods, thus allowing more efficient use of limited resources (Kumar and Kaushik, 2023). This expanded efficiency enables breeders to evaluate a broader range of genetic resources and extract greater biological value from germplasm collections (Kumar and Kaushik, 2023).

A notable example can be found in soybean research. Image-based phenotyping has been applied to assess diverse traits such as leaf area, biomass, senescence, nitrogen use efficiency, chlorophyll content, nutrient uptake, and yield estimation—all of which are key indicators of plant responses to abiotic stresses (Duc et al., 2023). Additionally, phenotypic features of soybean seeds—such as area, perimeter, length, and width—have been used to develop predictive models for seed weight based on image data, offering a non-destructive and scalable tool for breeding applications (Duc et al., 2023).

In maize, image-derived quantification of ear traits has been correlated with grain yield, enhancing the accuracy of yield prediction models (Resende et al., 2024). Similarly, in tomato, image-based techniques have been successfully employed to estimate individual fruit weight non-destructively (Farid et al., 2024). These examples illustrate the breadth of applications for image-based HTP across crop species and trait categories.

Beyond trait measurement, the integration of multidimensional phenotypic data with artificial intelligence and machine learning has paved the way for prescriptive phenotyping—an emerging concept that moves breeding from reactive selection to predictive intervention (Kumar and Kaushik, 2023). By anticipating plant responses under specific environmental scenarios, this approach enables the development of cultivars tailored to the precise needs of farmers and consumers. In this regard, HTP technologies mark a fundamental shift in breeding—transforming it from an empirical science into one of synthesis and predictive design.

A New Horizon for Breeding Advancement in Orphan Crops

Building on the advancements in HTP and standardized phenotyping, it is important to examine their transformative potential in under-researched contexts. Orphan crops, also referred to as neglected and underutilized species (NUS), play a crucial role in regional agriculture by offering climate resilience, rich nutritional profiles, and the presence of diverse functional compounds (Hossain et al., 2021; Tadele, 2019). Despite their ecological and agronomic importance, these crops have historically received limited scientific attention and investment compared to major staple crops (Akpojotor et al., 2025). Nonetheless, orphan crops are increasingly recognized as essential components of sustainable agriculture and food security strategies, especially in regions facing environmental stress and resource constraints.

Many of these crops require minimal irrigation, contribute to nitrogen fixation, enhance soil organic matter, and act as buffers against global threats such as climate change, emerging pests, and plant diseases (Akpojotor et al., 2025). As such, orphan crops are not merely substitutes for conventional staples but represent strategic opportunities to enhance the resilience and adaptability of future food systems (Kumar and Kaushik, 2023).

However, traditional breeding approaches remain prohibitively resource-intensive and time-consuming when applied to orphan crops, creating significant bottlenecks in their genetic improvement (Akpojotor et al., 2025). High-throughput phenotyping (HTP) offers a promising avenue to overcome these challenges by enabling precise and scalable trait measurements across thousands of plants in a relatively short period (Akpojotor et al., 2025). By utilizing images captured from ground-based platforms, unmanned aerial vehicles (UAVs), or satellite systems, breeders can generate high-quality phenotypic data for informed selection decisions—dramatically improving breeding efficiency, even in data-scarce environments (Akpojotor et al., 2025; Talabi et al., 2022).

In particular, UAV-based phenotyping has emerged as a cost-effective and accessible solution, making it especially attractive for orphan crop breeding programs in low-income countries (Resende et al., 2024). This shift enables breeding initiatives to move away from dependence on large labor forces and costly infrastructure, making research and varietal improvement for orphan crops not only feasible but scalable.

Examples underscore this shift. Table 2 summarizes key applications of HTP across orphan crops, highlighting the methods used, targeted traits, and resulting impacts. In papaya, image-based phenotyping has significantly reduced labor and time while improving the accuracy of yield-related trait evaluations (Cortes et al., 2018). In sorghum, more than 1,000 ear images from 272 genotypes were used to extract both one-dimensional and multidimensional phenotypic features, facilitating large-scale trait characterization and the development of genetic improvement strategies (Ibrahim Bio Yerima and Achigan-Dako, 2021). Similarly, chickpea breeding programs have leveraged multispectral and thermal imaging to assess disease progression in field conditions, providing real-time data for optimal intervention points in selection (Zhang et al., 2019). Buckwheat, another emerging orphan crop, has also been the focus of recent image-based phenotyping studies. By systematically collecting data on a wide range of traits, researchers have been able to more efficiently identify breeding targets and prioritize desirable characteristics for cultivar development (Oh et al., 2023). In lentils, HTP has enabled high-throughput screening for salinity tolerance, improving the efficiency of selection in stress-prone environments (Dissanayake et al., 2020). Additionally, in Camelina, the integration of multi-omics data with advanced phenotyping technologies has expedited the identification of stress-tolerant genotypes, further enhancing breeding pipelines (Großkinsky et al., 2023).

Collectively, these advancements demonstrate that HTP significantly reduces the time and cost traditionally associated with phenotypic data collection, effectively breaking through longstanding bottlenecks in orphan crop breeding (Chiurugwi et al., 2019; Tadele, 2019). By shifting the focus beyond major global staples, researchers and breeders are now able to harness a broader diversity of genetic resources and capitalize on the unique strengths of underutilized species. This not only diversifies the global food supply but also strengthens the overall resilience of agricultural systems in the face of climatic and demographic uncertainty.

Conclusion

Plant phenotyping has become an indispensable technology for understanding the complex interactions between genotype and environment, thereby enabling the development of improved crop varieties through advanced breeding strategies. By overcoming the limitations of traditional manual methods, phenotyping—particularly when integrated with high-throughput approaches—has played a pivotal role in resolving bottlenecks in the breeding pipeline. It has further evolved toward prescriptive phenotyping, where predictive analytics guide trait selection based on environmental scenarios and agronomic needs.

A key foundation for such progress lies in the standardization of genetic resource data collection. Standardized protocols and ontologies enhance interoperability across studies, facilitate large-scale comparative analyses, and lay the groundwork for precision breeding. In this context, high-throughput phenotyping (HTP) is not only accelerating breeding in major crops but is also opening new horizons for the research and improvement of orphan and underutilized crops, which have historically received limited scientific attention despite their agronomic and nutritional value.

However, several challenges remain in the widespread implementation of HTP and standardization practices. These include the high costs associated with stationary phenotyping platforms, the complexity of integrating and managing heterogeneous sensor data, and the lack of standardized formats and data-sharing infrastructures (Gill et al., 2022). Addressing these obstacles will require the systematic adoption of the FAIR data principles—Findable, Accessible, Interoperable, and Reusable—which provide a robust framework for data governance and reuse in agricultural research (Ali and Dahlhaus, 2022; Gill et al., 2022).

The implementation of FAIR data principles will not only enhance data reuse but also enable international collaborations and meta-analyses that accelerate genetic discovery across multiple crops and environments.

Ultimately, with the support of HTP technologies and standardized phenotypic frameworks, it will become increasingly feasible to identify desirable traits across diverse crop species, respond proactively to environmental change, and even anticipate future agricultural challenges. The strategic deployment of these innovative tools will be central to advancing the future of precision agriculture and ensuring the long-term sustainability of global food production.