Predicting Plant Growth and Development Using Time-Series Images

Early prediction of the growth and development plants is important for intelligent breeding process, yet accurate simulation plant phenotypes difficult. In this work, a model based on spatiotemporal long short-term memory (ST-LSTM) in network (MIM) was proposed to predict image sequences future including organs such as ears. A novel dataset wheat also compiled. The performance evaluated by calculating structural similarity index measure (SSIM), mean square error (MSE), peak signal noise ratio (PSNR) between predicted real images. Moreover, optimal number time steps interval were determined dataset. Under setting, SSIM values surpassed 84% all steps. MSE 46.11 below 68 PSNR 30.67. When set eight, had best public Panicoid Phenomap-1 78% 77.78 118 29.03. results showed high degree images verified validity, reliability, feasibility model. study shows potential provide phenotyping community with an efficient tool that can perform high-throughput growth.