For efficient design of post-harvest processing machines/equipment, knowledge of the physical properties of agricultural materials is required. Individual measurements of respective material’s properties are typically used to calculate the equipment/machines design parameters. These properties are important for grading and sorting processes. This study experimentally measured the individual dimensional properties of the seeds and by simulation used artificial neural network to predict some physical properties of African locust bean (Parkia biglobosa) seeds. Two artificial neural network structures, the Multilayer Propagation Neural Network (MLP) and the Radial Basis Neural Network (RBF) were employed. The dimensional properties of the seeds were used as the input variables and the derived seeds' properties - volume, surface and projected regions, geometric mean diameter and sphericity - were used as the output variables. The results showed that MLP structures of the artificial neural network could provide a better prediction of the seeds physical properties parameters, making it a viable replacement method for the manual approach.