Fig. 15

Comparison of classifier cross validation accuracy as a function of the input feature set used. One can observe that the multi-modal nature of the dataset provides the highest overall classifier performance. One can also observe how different kinds on features influence the prediction performance for different stress factors, e.g. the 3D data is an important indicator of weed pressure due to the significant height difference between the plants and weeds, resulting in a lower mean height while increasing the variance compared to the control group