We present a systematic evaluation of low-cost soil sensors for early stress and disease detection in avocado plants. Our monitoring system was deployed across 72 plants divided into four treatment categories within a controlled greenhouse environment collecting data over six months. We developed a two-level hierarchical classifier leveraging soil electrical conductivity (EC) and moisture data to improve classification accuracy. The proposed classifier achieved 75--86% accuracy across different avocado genotypes, outperforming conventional machine learning approaches by over 20%. Our findings demonstrate that while low-cost sensors exhibit certain limitations in field conditions, strategic classification techniques can significantly enhance their utility for precision agriculture.

With rising demands for efficient disease and salinity management in agriculture, early detection of plant stressors is crucial, particularly for high-value crops like avocados. This paper presents a comprehensive evaluation of low-cost sensors deployed in the field for early stress and disease detection in avocado plants. Our monitoring system was deployed across 72 plants divided into four treatment categories within a greenhouse environment, with data collected over six months. While leaf temperature and conductivity measurements, widely used metrics for controlled settings, were found unreliable in field conditions due to environmental interference and positioning challenges, leaf spectral measurements produced statistically significant results when combined with our machine learning approach. For soil data analysis, we developed a two-level hierarchical classifier that leverages domain knowledge about treatment characteristics, achieving 75-86% accuracy across different avocado genotypes and outperforming conventional machine learning approaches by over 20%. In addition, performance evaluation on an embedded edge device demonstrated the viability of our approach for resource-constrained environments, with reasonable computational efficiency while maintaining high classification accuracy. Our work bridges the gap between theoretical potential and practical application of low-cost sensors in agriculture and offers insights for developing affordable, scalable monitoring systems.

Recent advances in Internet-of-Things (IoT) technologies have sparked significant interest towards developing learning-based sensing applications on embedded edge devices. These efforts, however, are being challenged by the complexities of adapting to unforeseen conditions in an open-world environment, mainly due to the intensive computational and energy demands exceeding the capabilities of edge devices. In this paper, we propose OpenSense, an open-world time-series sensing framework for making inferences from time-series sensor data and achieving incremental learning on an embedded edge device with limited resources. The proposed framework is able to achieve two essential tasks, inference and incremental learning, eliminating the necessity for powerful cloud servers. In addition, to secure enough time for incremental learning and reduce energy consumption, we need to schedule sensing activities without missing any events in the environment. Therefore, we propose two dynamic sensor scheduling techniques: (i) a class-level period assignment scheduler that finds an appropriate sensing period for each inferred class, and (ii) a Q-learning-based scheduler that dynamically determines the sensing interval for each classification moment by learning the patterns of event classes. With this framework, we discuss the design choices made to ensure satisfactory learning performance and efficient resource usage. Experimental results demonstrate the ability of the system to incrementally adapt to unforeseen conditions and to efficiently schedule to run on a resource-constrained device.

The rapid advancement of IoT technologies has generated much interest in the development of learning-based sensing applications on embedded edge devices. However, these efforts are being challenged by the need to adapt to unforeseen conditions in an open-world environment. Updating a learning model suffers from the lack of training data as well as the high computational demand beyond that available on edge devices. In this paper, we propose an open-world time-series sensing framework for making inferences from time-series sensor data and achieving incremental learning on an embedded edge device with limited resources. The proposed framework is able to achieve two essential tasks, inference and learning, without requiring access to a powerful cloud server. We discuss the design choices made to ensure satisfactory learning performance and efficient resource usage. Experimental results demonstrate the ability of the system to incrementally adapt to unforeseen conditions and to effectively run on a resource-constrained device.