Comparative Analysis of High‐Throughput Phenotyping Techniques for Early Detection of Herbicide Stress

Early detection of herbicide‐induced stress is essential for optimising weed control, improving crop safety and advancing precision agriculture practices. While non‐destructive imaging technologies offer great potential for rapid stress diagnosis, direct comparisons of their performance across species, herbicides and application rates remain scarce. This study systematically compared three high‐throughput imaging methods: chlorophyll fluorescence imaging, multispectral imaging and 3D multispectral scanning, for their ability to detect and classify early physiological and morphological responses to bentazone (HRAC 6) and glyphosate (HRAC 9) in a weed (velvetleaf, Abutilon theophrasti ) and a crop species (common bean, Phaseolus vulgaris ). Plants were imaged daily, immediately before treatment and for five consecutive days after herbicide application. Chlorophyll fluorescence parameters, particularly NPQ, q P and F s ′, emerged as the earliest and most sensitive indicators, detecting stress within 24 h of treatment. Stepwise discriminant analysis revealed that intermediate time points (48–72 h after application) achieved the highest classification accuracies, reaching up to 100%, with chlorophyll fluorescence as the dominant trait for selection. Multispectral traits, such as reflection in red and far red, and saturation, together with morphological traits like total leaf area and leaf inclination, provided complementary information and enhanced discrimination accuracy among herbicides. These findings highlight the superiority of chlorophyll fluorescence imaging for detecting early herbicide stress, while demonstrating the added value of integrating spectral and morphological traits. The results support the development of multi‐sensor phenotyping pipelines for rapid, accurate and field‐adaptable diagnostics in both weed and crop management.