By use of experimental data, gathered from a pilot-scale rotary drum, which was made from transparent plastic materials, an improved model was developed for describing the steady state bed depth profile of material flow in a rotary drum. The structure of the improved model is based on the existing steady state model with four added parameters. Development of the improved model includes two stages. At the first stage, the boundary condition of the original model is corrected by use of the experimental data gathered from the bed depth of materials at the discharge cross section with Support Vector Regression (SVR) method. Then, the parameters of the improved model were calculated, using the bed depth of materials at various locations along the drum with the Genetic Algorithm (GA) method. Image processing techniques were used for measurement of the bed depth at different locations along the drum. The improved model was determined to have better fitness at all the ranges of operating and design parameters, particularly, at high values of the slope of drum and low values of rotational speed as well as the discharge end region, where the original model is not working well. The Mean Square Errors (MSE) of the original and the improved models were 0.122 and 0.014, respectively.