Succulent plants survive in arid environments by storing water and nutrients in their leaves, stems, and roots, forming thick tissues. ice plants, a type of succulent, are well known for their benefits such as blood sugar regulation and antioxidant properties. however, they exhibit low germination rates and non-uniform growth, leading to high management costs during cultivation. to address this issue, early identification of plants with potential non-uniform growth in the initial sowing stage is crucial to improving cultivation efficiency. this study proposes a deep learning-based approach utilizing object segmentation and time-series prediction models to estimate the growth of ice plants quantitatively. we collected videos over 17 days in a sealed thin-film hydroponic system and extracted crop images at different growth stages. The YOLOv8n-seg model was applied to segment individual crops at the pixel level, allowing the calculation of area-based growth information. Additionally, a Long ShortTerm Memory (LSTM) model was used to analyze time-sequenced area data, predicting the growth trends of 54 individual ice plants, and the results were aggregated for further analysis. this study serves as a fundamental analysis of ice plant growth patterns using time-seriesbased area data, providing essential insights for developing an automated crop growth monitoring system in smart farming environments. furthermore, the proposed method can be extended to other crops with similar growth characteristics, contributing to advancements in precision agriculture and smart farming technologies.