The dataset contains two folders: 1. training_sample 46 samples belonging to 16 object classes, stored in xyzl format. 2. testing_scene 3 piece of data collected in PKU campus, and 9 piece of data collected in ShangDi street. All stored in xyzl format too. An example of .xyzl data: #x y z(in world coord) label 1437.897 26289.970 -89.238 14 #3 points labeled as building 1436.272 26288.056 -89.412 14 1436.145 26287.906 -89.685 14 1428.232 26278.583 -89.113 0 #3 unlabeled points 1428.301 26278.665 -89.297 0 1428.414 26278.798 -89.490 0 1417.659 26266.128 -88.233 1 #5 points labeled as car 1417.618 26266.079 -88.257 1 1417.596 26266.053 -88.284 1 1417.592 26266.049 -88.316 1 1417.621 26266.082 -88.356 1 The object label of laser point used in the dataset: undefine 0 car 1 bus 2 truck 3 bicycle 4 pedestrian 5 pole 6 traffic light 7 road lamp 8 road sign 9 traffic sign 10 trash box 11 curb 12 pile 13 building 14 fence 15 adboard 16