A universal problem to all LiDAR devices is that cars with dark paint have a very low optical reflectivity (4%) causing them to be much harder to detect than for example a white car (62%). While improvements are being made by developing special LiDAR friendly paint, detecting black "legacy" cars in LiDAR data is still very relevant. Typically, any car will have areas of higher reflectivity like license plates, lamps and rims which are very visible to the LiDAR. The scope of this project is to research how such cars could be detected in LiDAR data using either classic approaches or machine learning. At least to a human, the lack of return points in the shape of a car still allows us to "see" the cars despite them missing from the data entirely. Unfortunately, we cannot offer compensation for work on these theses.