Computer-implemented automatic annotation of a LiDAR point cloud
Examples are disclosed that relate an approach for automatically annotating a LiDAR point cloud of an airport with classification labels. In one example, a computing device comprises one or more processors configured to execute instructions stored in memory to receive LiDAR data for an airport environment, wherein the LiDAR data includes a LiDAR point cloud including a plurality of points in the airport environment, segment the LiDAR point cloud into a ground region and a non-ground region, automatically cluster, via a trained machine learning model, different points in the non-ground region into a plurality of non-ground segments, automatically annotate, via the trained machine learning model, the plurality of non-ground segments with classification labels selected from an airport domain, and output a panoptic labeled LiDAR point cloud including the plurality of non-ground segments labeled with respective classification labels and the points in the ground region.
1 . A computing device, comprising:
one or more processors configured to execute instructions stored in memory to:
receive laser imaging detection and ranging (LiDAR) data for an airport environment, wherein the LiDAR data includes a LiDAR point cloud including a plurality of points in the airport environment;
segment the LiDAR point cloud into a ground region and a non-ground region;
automatically cluster, via a trained machine learning model, different points in the non-ground region into a plurality of non-ground segments;
automatically annotate, via the trained machine learning model, the plurality of non-ground segments with classification labels selected from an airport domain, wherein the airport domain includes a library of classification labels corresponding to airport-related objects; and
output a labeled LiDAR point cloud including the plurality of non-ground segments labeled with respective classification labels and the points in the ground region.
2 . The computing device of claim 1 , wherein the LiDAR point cloud includes points in the airport environment accumulated from a plurality of scans of one or more LiDAR devices.
3 . The computing device of claim 1 , wherein at least some of the LiDAR data is received from one or more aircraft-mounted LiDAR devices.
4 . The computing device of claim 1 , wherein each point included in the LiDAR point cloud includes X, Y, and Z coordinates and an intensity value.
5 . The computing device of claim 1 , wherein the classification labels include dynamic semantic labels for movable objects and non-dynamic semantic labels for fixed objects.
6 . The computing device of claim 1 , wherein the one or more processors are configured to execute instructions stored in memory to generate a plurality of masks corresponding to the plurality of non-ground segments, and wherein the labeled LiDAR point cloud includes the plurality of masks.
7 . The computing device of claim 1 , wherein the one or more processors are configured to execute instructions stored in memory to:
annotate each of the plurality of non-ground segments with an individualized identifier.
8 . The computing device of claim 1 , wherein the one or more processors are configured to execute instructions stored in memory to:
receive one or more images of the airport environment captured by one or more cameras; and
automatically annotate a plurality of objects in the one or more images with classification labels corresponding to the plurality of non-ground segments in the labeled LiDAR point cloud.
9 . The computing device of claim 1 , wherein the labeled LiDAR point cloud is included in a set of training data, and wherein the one or more processors are configured to execute instructions stored in memory to re-train the trained machine learning model based at least on the set of training data.
10 . A computer-implemented method comprising:
receiving LiDAR data for an airport environment, wherein the LiDAR data includes a LiDAR point cloud including a plurality of points in the airport environment;
segmenting the LiDAR point cloud into a ground region and a non-ground region;
automatically clustering, via a trained machine learning model, different points in the non-ground region into a plurality of non-ground segments;
automatically labeling, via the trained machine learning model, the plurality of non-ground segments with classification labels selected from an airport domain, wherein the classification labels include dynamic semantic labels for movable objects and non-dynamic semantic labels for fixed objects; and
outputting a labeled LiDAR point cloud including the plurality of non-ground segments labeled with respective classification labels and the points in the ground region.
11 . The computer-implemented method of claim 10 , further comprising:
generating a plurality of masks corresponding to the plurality of non-ground segments, and wherein the labeled LiDAR point cloud includes the plurality of masks.
12 . The computer-implemented method of claim 10 , further comprising:
annotating via the trained machine learning model, each of the plurality of non-ground segments with an individualized identifier.
13 . The computer-implemented method of claim 10 , further comprising:
receiving one or more images of the airport environment captured by one or more cameras; and
automatically annotating a plurality of objects in the one or more images with classification labels corresponding to the plurality of non-ground segments in the labeled LiDAR point cloud.
14 . The computer-implemented method of claim 10 , wherein the labeled LiDAR point cloud is included in a set of training data, and wherein the computer-implemented method further comprises re-training the trained machine learning model based at least on the set of training data.
15 . The computer-implemented method of claim 10 , wherein the LiDAR point cloud includes points in the airport environment accumulated from a plurality of scans of one or more LiDAR devices.
16 . The computer-implemented method of claim 10 , wherein at least some of the LiDAR data is received from one or more aircraft-mounted LiDAR devices.
17 . The computer-implemented method of claim 10 , wherein the airport domain includes a library of classification labels corresponding to airport-related objects.
18 . A computing device, comprising:
one or more processors configured to execute instructions stored in memory to:
receive laser imaging detection and ranging (LiDAR) data for an airport environment, wherein the LiDAR data includes a LiDAR point cloud including a plurality of points in the airport environment;
segment the LiDAR point cloud into a ground region and a non-ground region;
automatically cluster, via a trained machine learning model, different points in the non-ground region into a plurality of non-ground segments;
automatically annotate, via the trained machine learning model, the plurality of non-ground segments with classification labels selected from an airport domain;
output a labeled LiDAR point cloud including the plurality of non-ground segments labeled with respective classification labels and the points in the ground region;
receive one or more images of the airport environment captured by one or more cameras; and
automatically annotate a plurality of objects in the one or more images with classification labels corresponding to the plurality of non-ground segments in the labeled LiDAR point cloud.
19 . The computing device of claim 18 , wherein the one or more processors are configured to execute instructions stored in memory to:
automatically annotate each of the plurality of non-ground segments with an individualized identifier.
20 . The computing device of claim 18 , wherein the classification labels include dynamic semantic labels for movable objects and non-dynamic semantic labels for fixed objects.