Biology subjects

de Jesus Colwell, F.

Publications and source records attributed to de Jesus Colwell, F..

3 recordsLinked to original sources

bioRxiv2025

Synergistic 3D, multispectral, and thermal image analysis via supervised machine learning for improved detection of root rot symptoms in hydroponically-grown flat-leaf parsley

Root rot in hydroponically-grown leafy vegetables is difficult to detect via conventional manual and machine vision-based approaches as symptoms of infection are not clearly visible on the canopy at earlier stages of infection. Hence, the present study investigates the potential of using machine learning for assessing canopy information obtained from multiple imaging platforms synergistically to improve root rot detection. Herein, flat-leaf parsley seedlings were grown in an experimental hydroponic vertical farm and inoculated with Pythium irregulare and Phytophthora nicotianae. Subsequently, the seedlings were imaged via 3D, multispectral, and thermal sensors at various stages of growth to obtain twenty-six image-based plant features. Following a preliminary screening of redundant features via regression analysis, data for seventeen image features associated with morphometric, spectral, and thermal attributes was co-analyzed using supervised machine learning by Support Vector Machines (SVM). Exhaustive feature selection using different SVM kernels and maximum feature thresholds was performed to identify optimal feature subsets. It was observed that combining parameters obtained from all three imaging platforms enabled better identification of infected samples (>99%) than using a higher number of attributes from individual imaging systems. In addition, model performance was improved considerably by including temporal information during model training. Hence, it may be inferred that fusion of data from multiple imaging systems and using it with temporal information can enable better real-time high-throughput monitoring of root rot.

plant biology↗

bioRxiv2024

Infrared thermography for plant stress detection in vertical farms: Investigating spatiotemporal variations and exploring solutions via machine learning

Application of infrared thermography (IRT) for real-time plant stress detection has grown rapidly in recent years. Although the technology has been well established for crops grown in fields and glasshouses, its feasibility for vertical farms has not been tested extensively. In this study, temporal monitoring of stress induced by root dehydration in purple basil plantlets inside a vertical farm was performed to identify bottlenecks in real-time stress detection via IRT. Subsequently, potential solutions were investigated via machine learning by implementing support vector machines for supervised classification. Edge effects as well as proximity to air vents were identified as the major causes of positional variation in plant temperature that could lead to misprediction of stress. Binary, ternary, and quaternary classification models were trained using thermal images from two, three, and four levels of stress, respectively, to assess model performance. Binary classification models trained with plants experiencing medial and high levels of stress were able to identify stressed plants with high accuracy (81-94%). Further, binary models trained using plants under medial levels of stress generated a continuous probability distribution for stress prediction when plotted against plant temperature. In contrast, models trained using samples experiencing high stress generated distinct probabilistic clusters for the unstressed and highly stressed plants, but were unable to classify medial stress samples reliably. Similarly, ternary and quaternary models were able to better predict very high and very low levels of stress than intermediate stress levels. Hence, our findings suggest that binary classification models trained using samples under medial levels of stress would be helpful in overcoming spatiotemporal variations in canopy thermal profile by providing reliable probabilistic estimates of plant stress within a vertical farming system. Key pointsO_LIPlant stress detection in vertical farms via thermal imaging may be challenging because perceptible plant temperature can be strongly influenced by its microenvironment. C_LIO_LIThermal image analysis via supervised machine learning allows the development of robust prediction models that can overcome such factors to identify stressed plants. C_LIO_LIBinary classification machine learning models can reliably identify stressed plants as well as provide probabilistic estimates for the degree of stress. C_LI

plant biology↗

bioRxiv2024

Assessing nutritional pigment content and plant health status of green and red leafy vegetables by image analysis: Catching the 'red herring' of plant digital color processing

Estimating pigment content in leafy vegetables via digital image analysis is a reliable method for real-time assessment of plant health status and nutritive value. However, the present leaf color analysis models developed using green-leafed plants do not perform reliably while analyzing images of anthocyanin-rich leaves, often giving misleading or "red herring" trends. Hence, the present study investigates variations in different digital color features for six types of leafy vegetables with varying levels of laminar anthocyanin to identify holistic digital color analysis models that could be implemented for real-time assessment of health status as well as nutritional pigment contents of leafy vegetables irrespective of laminar anthocyanin status. For this, datasets from three digital color spaces, viz., RGB (red, green, blue), HSI (hue, saturation, intensity), and L*a*b* (lightness, redness-greenness, yellowness-blueness), were compared with pigment contents of n = 320 leaf samples, and were analyzed via linear and non-linear regression using single variables, multiple linear regression, Support Vector regression, and Random Forest regression to predict chlorophyll and anthocyanin contents. While most digital color features presented abrupt shifts between anthocyanin-rich and low-anthocyanin samples, the R digital color feature did not show any deviation due to leaf anthocyanin content and was found to have the best correlation with SPAD chlorophyll meter readings (R2 = 0.83). Concomitantly, H (R2 = 0.82) and a* (R2 = 0.79) features correlated most strongly with leaf anthocyanin content. In general, prediction of pigment contents was more accurate when data from all channels within a color space was analyzed simultaneously. Further, most reliable estimates of pigment content were provided by Support Vector and Random Forest regression models (0.7 < R2 < 0.85). Thus, the present findings demonstrate how digital color analysis of green as well as anthocyanin-rich leafy vegetables could be implemented for assessing plant health status and nutritional pigment content non-invasively.

plant biology↗