Application of RGB-Imaging techniques for high-throughput plant phenotyping- A Review

Abstract

High-throughput plant phenotyping plays an important role in plant breeding, identifying superior genotypes in a fast and accurate manner. Consequently, researchers are seeking new image-based plant phenotyping strategies to enhance phenotyping efficiency. RGB (red, green, blue) sensors and their applications offer a cost-effective and accessible approach while maintaining the advantageous characteristics of high-throughput phenotyping technologies. Various RGB image-based indices are now available to measure diverse phenotypic traits accurately. Despite a wide range of advantages, some limitations reduce the accuracy of the method. We systematically reviewed scientific articles published between 2000 and 2024 to ascertain the significance, available knowledge, and gaps in the domain of RGB image-based plant phenotyping. This review paper provides a comprehensive survey on the significance of current RGB imaging technologies and their applications in plant phenotyping, emphasizing their advantages and limitations. RGB image-based plant phenotyping demonstrates considerable accuracy in the estimation of morphological traits. However, this technique gives significantly lower accuracy for physiological traits compared with other sensors. Furthermore, variation of light conditions, varied backgrounds, and overlapping are major drawbacks of this technique. Future studies should focus on the development of precise image acquisition systems, advanced image processing techniques (including image segmentation), the identification of novel color parameters, the implementation of robust artificial intelligence and machine learning models, and the integration of complementary sensor technologies to address existing challenges.

Code and data availability

This is a systematic review of RGB image-based plant phenotyping literature (2000–2024). The article contains no paper-specific phenotype datasets, plant images, sensor data, analysis code, or trained models; all cited works are prior publications, and no data or code availability statement appears in the supplied text

No evidence-backed public reproduction asset is currently recorded.