Smartphones show potential for controlling and monitoring variables in agriculture. Their processing capacity, instrumentation, connectivity, low cost, and accessibility allow farmers (among other users in rural areas) to operate them easily with applications adjusted to their specific needs. In this investigation, the integration of inertial sensors, a GPS, and a camera are presented for the monitoring of a coffee crop. An Android-based application was developed with two operating modes: (i) Navigation: for georeferencing trees, which can be as close as 0.5 m from each other; and (ii) Acquisition: control of video acquisition, based on the movement of the mobile device over a branch, and measurement of image quality, using clarity indexes to select the most appropriate frames for application in future processes. The integration of inertial sensors in navigation mode, shows a mean relative error of ±0.15 m, and total error ±5.15 m. In acquisition mode, the system correctly identifies the beginning and end of mobile phone movement in 99% of cases, and image quality is determined by means of a sharpness factor which measures blurriness. With the developed system, it will be possible to obtain georeferenced information about coffee trees, such as their production, nutritional state, and presence of plagues or diseases.

Recent years have seen significant technological advancements in precision farming and plant phenotyping. Remote sensing along with deep learning (DL) techniques can increase phenotyping efficiency and help on-farm decision making with rapid stress detection. In this work, we use these techniques to evaluate drought stress in soybean plants, a crop whose yield is significantly affected by water availability. Images were taken from a high vantage in the field at various times throughout the day. Each image is given a wilting score ranging from 0 to 4 by expert scorers. We implement a DL method called multiple instance learning (MIL) to perform wilt classification as well as generate heat maps that highlight wilt levels in specific regions of the image. Given the significant overlap between adjacent classes in our dataset, we were able to achieve an overall classification accuracy of 64% and a one-off accuracy of 96% on our holdout test set. Our model outperformed DenseNet121 in most metrics, and provided comparable performance to a vision transformer (ViT) while having fewer parameters overall, less complexity (useful for edge implementations), and some interpretability. Furthermore, we were able to show that our model outperformed expert human annotators by predicting more consistent and accurate wilt levels when considering single-image re-annotation. The results show that our proposed methodology can be a useful approach in detecting drought stress in soybean fields to facilitate efficient crop management and aid selection of drought-resilient varieties.