Computer Vision and Machine Learning-Based Predictive Analysis for Urban Agricultural Systems

Urban Farm DB summary

A study on microclimate monitoring for urban agriculture (region unspecified) evaluated convolutional neural networks (CNNs) forecasting relative air humidity, soil moisture, and light intensity sensor readings from thermal images. Forecasts for relative humidity and soil moisture achieved higher accuracy, with Mean Absolute Percentage Errors of 10-12%, attributed to their strong dependency on thermal patterns captured by the CNNs, while light intensity forecasting was less accurate due to more complex and variable urban factors. The authors acknowledge that implementing these technologies in urban agricultural models still poses challenges.