Camera parameter estimation using semantic labels

A device obtains a point cloud of a scene including a plurality of points. Each point has three-dimensional coordinates in a three-dimensional coordinate system. A first cluster of points has a first semantic label. The device obtains a two-dimensional image of the scene with a camera with an intrinsic parameter. The device detects, in the two-dimensional image, a representation of a first object corresponding to the first semantic label. The device determines two-dimensional coordinates in a two-dimensional coordinate system of the two-dimensional image corresponding to the first object. The device determines, from the first cluster of points, three-dimensional coordinates in the three-dimensional coordinate system of the scene corresponding to the two-dimensional coordinates in the two-dimensional coordinate system of the two-dimensional image of the scene. The device estimates the intrinsic parameter based on the two-dimensional and the three-dimensional coordinates.