Train with own dataset

[jun-ja]

Hi, I'm trying to train with Seeing Through Fog(STF) dataset (https://github.com/princeton-computational-imaging/SeeingThroughFog).

Since the STF dataset follows the kitti label format, it was easy to change to the kitti format. However, there is a slight difference in class name, so I don't know how to change it. I tried to change it appropriately, but the category of the kitti label class was more subdivided than the category of the STF label. STF's label consists of 'PassengerCar', 'LargeVehicle', 'Pedestrian', 'RidableVehicle', 'Dontcare', and each detailed label. How do I change these to kitti's class names (Car, Van, Truck, Pedestrian, Person_sitting, Cyclist, Tram, Misc, or DontCare)?