Several analytical as well as numerical models have been proposed over the years for the estimation of velocity distribution in vegetated channels. However, in the case of flexible submerged aquatic vegetations, the analysis becomes complex due to the introduction of variable flexural rigidity (EI) and deflected stem height (hv). To compute these two parameters, experiments were carried out with live aquatic plants of three species (water lily, water chestnut, and lotus) to determine the variable flexural rigidity (using own-weight cantilever method) and deflected stem height (open channel flume test).To make the computational effort even more straightforward, a feed-forward artificial neural network (ANN) model was trained [with only 5 input parameters and velocity distribution in the vegetation layer as output] to predict the velocity distribution. The model performed with a satisfactory R2 value (0.97 for training, 0.97 for validation, and 0.96 for testing) showing its applicability as a supplement to the existing velocity distribution models. Finally, the feature importance of the model showed that the flexural rigidity of the plant stem is in fact the most significant parameter in the prediction of velocity distribution thus requiring careful estimation rather than gross approximation as a constant parameter for plant species as assumed by past research works. The motivation of the present study lies in reducing the computational effort required in analytical velocity models through the ANN model while considering the spatial variability of flexural rigidity of plant stems.