Crop growth monitoring with time series data based on deep learning

This study presents a model based on deep learning to analyze the interaction of variables that influence plant growth in greenhouse crops. To achieve this, a multi-category database has been collected in two types of open and semi-open greenhouses throughout the year during three growing seasons, including two cultivars of tomato plants with a sample of 100 plants per season. The database consists of images captured at the growth point of the plants, phenotyping data obtained weekly, and environmental data with a sampling frequency per minute. The implemented model uses time series data to find the relationship of variables and automatically predict the state of the crop at the individual plant level and further extended as a group of plants. In summary, the proposed research shows a practical solution to monitor plants over time and provides strategies for multi-category data collection that allow a better understanding of the dynamics of plant stress factors.