>Perhaps most egregiously, 217 (1.4%) of the images were completely unlabeled but actually contained cars, trucks, street lights, and/or pedestrians.
This gives us a data set size of 15500.
>Perhaps most egregiously, 217 (1.4%) of the images were completely unlabeled but actually contained cars, trucks, street lights, and/or pedestrians.
This gives us a data set size of 15500.