Modern phenotyping and plant disease detection provide promising step towards food security and sustainable agriculture. In particular, imaging and computer vision based phenotyping offers the ability to study quantitative plant physiology. On the contrary, manual interpretation requires tremendous amount of work, expertise in plant diseases, and also requires excessive processing time. In this work, we present an approach that integrates image processing and machine learning to allow diagnosing diseases from leaf images. This automated method classifies diseases (or absence thereof) on potato plants from a publicly available plant image database called 'Plant Village'. Our segmentation approach and utilization of support vector machine demonstrate disease classification over 300 images with an accuracy of 95%. Thus, the proposed approach presents a path toward automated plant diseases diagnosis on a massive scale.

No abstract

Avocado, a climacteric fruit, exerts high rate of respiration and ethylene production and thereby subject to ripening during storage. erefore, its ripening is a signi cant factor to impart optimum quality in postharvest storage. To understand the dynamics of ripening and to assess the degree of ripening in the avocado, electrical sensing technique is utilized in this study. In particular, electrical impedance spectroscopy (EIS) is found to uncover the physiological and structural characteristics in plants and vegetables and to follow physiological progressions due to environmental impacts. In this work, we present an approach that will integrate EIS and machine learning technique that allows us to monitor the ripening degree of the avocado. It is evident from our study that the impedance absolute magnitude of the avocado gradually decreases as the ripening stages ( rm, breaking, ripe, and overripe) proceed at a particular frequency. In addition, principal component analysis shows that impedance magnitude (two principal components combined explain 99.95% variation) has better discrimination capabilities for ripening degrees compared to impedance phase angle, impedance real part, and impedance imaginary part. Our classi er utilizes two principal component features over 100 EIS responses and demonstrates classi cation over rm, breaking, ripe, and overripe stages with an accuracy of 90%, precision of 93%, recall of 90%, f1-score of 90%, and auc of 88%. e study o ers plant scientists a low cost and nondestructive approach to monitor postharvest ripening process for quality control during storage.

Imaging and computer vision systems offer the ability to study quantitatively on human physiology. On contrary, manual interpretation requires tremendous amount of work, expertise and excessive processing time. This work presents an algorithm that integrates image processing and machine learning to diagnose diabetic retinopathy from retinal fundus images. This automated method classifies diabetic retinopathy (or absence thereof) based on a dataset collected from some publicly available database such as DRIDB0, DRIDB1, MESSIDOR, STARE and HRF. Our approach utilizes bag of words model with Speeded Up Robust Features and demonstrate classification over 180 fundus images containing lesions (hard exudates, soft exudates, microaneurysms, and haemorrhages) and non-lesions with an accuracy of 94.4%, precision of 94%, recall and f1-score of 94% and AUC of 95%. Thus, the proposed approach presents a path toward precise and automated diabetic retinopathy diagnosis on a massive scale.

Onion is perishable and thereby subject to drying during unrefrigerated storage. Its moisture content is important to ensure optimum quality in storage. To track and analyze the dynamics of natural dehydration in onion and also to assess its moisture content, noninvasive and nondestructive methods are preferred. One of them is known as electrical impedance spectroscopy (or EIS in short). In the first phase of our experiment, we have used EIS, where we apply alternating current with multiple frequency to the object (onion in this case) and generate impedance spectrum which is used to characterize the object. We then develop an equivalent electrical circuit representing onion characteristics using a computer assisted optimization technique that allows us to monitor the response of onion undergoing natural drying for a duration of 3 weeks. The developed electrical model shows better congruence with the impedance data measured experimentally when compared to other conventional models for plant tissue with a mean absolute error of 0.42% and root mean squared error of 0.55%. In the second phase of our experiment, we attempted to find a correlation between the previous impedance data and the actual moisture content of the onions under test (measured by weighing) and developed a mathematical model. This model will provide an alternative tool for assessing the moisture content of onion nondestructively. Our model shows excellent correlation with the ground truth data with a deterministic coefficient of 0.9767, root mean square error of 0.02976 and sum of squared error of 0.01329. Therefore, our two models will offer plant scientists the ability to study the physiological status of onion both qualitatively and quantitatively.