The result of this research, the river depth predicting NN, could eventually have applications in multiple different areas. One of these areas is safety during a flood. With climate change, flooding will become a more apparent issue. Floods will become more extreme and more common. If an app exists that can tell with certainty if a river or stream is safe to cross, it could lower the number of deaths caused by drowning during floods.
Estimating river depth using AI is still in its infancy, however, it has a future. In this paper, it was found that a Convolutional Neural Network is the best type to use for the purposes of this research and finally it was found that neural networks are able to predict a clear river’s depth and that they do this with an accuracy of 17%.
This research implies that the used theory works. The accuracy was higher than the accuracy if the NN guessed randomly, which suggests that some pattern was found. Perhaps with a larger dataset, overfitting as mentioned in can be avoided and produce a better accuracy closer to the hypothesis. This potential can be explored in further research.