Method and apparatus for processing point cloud data

Patent metadata

Field	Value
Publication number	US-10268926-B2
Application number	US-201715410713-A
Country	US
Kind code	B2
Filing date	Jan 19, 2017
Priority date	Sep 27, 2016
Publication date	Apr 23, 2019
Grant date	Apr 23, 2019

Abstract

The present application discloses a method and an apparatus for processing point cloud data. The method in an embodiment comprises: presenting a to-be-labeled point cloud frame and a camera image formed by photographing an identical scene at an identical moment as the point cloud frame; determining, in response to an operation of selecting an object in the point cloud frame by a user, an area encompassing the selected object in the point cloud frame; projecting the area from the point cloud frame to the camera image, to obtain a projected area in the camera image; and adding a mark in the projected area, for labeling, by the user, the selected object in the point cloud frame according to the mark indicating an object in the camera image. This implementation can assist labeling personnel in rapidly and correctly labeling an object in a point cloud frame.

First claim

What is claimed is:

1. A method for processing point cloud data, comprising:
displaying a to-be-labeled point cloud frame within an interface;
displaying a camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface, the to-be-labeled point cloud frame being non-superimposed to the camera image;
determining, in response to an operation of selecting a displayed object in the point cloud frame by a user, an area encompassing the selected object in the point cloud frame;
projecting the area from the point cloud frame to the camera image, to obtain a projected area in the camera image;
displaying a mark in the projected area, for labeling, by the user, the selected object in the point cloud frame according to the mark indicating an object in the camera image; and
displaying simultaneously (i) the to-be-labeled point cloud frame, (ii) the camera image, and (iii) label options for labeling the displayed object in the to-be-labeled point cloud frame within the interface, wherein the displayed to-be-labeled point cloud frame, the displayed camera image and the displayed label options are non-superimposed to each other.

2. The method according to claim 1 , further comprising:
generating, in response to the labeling, by the user, the selected object in the point cloud frame according to the mark, a labeled point cloud frame according to operation information of the labeling.

3. The method according to claim 2 , further comprising:
training an obstacle recognition algorithm by using the labeled point cloud frame as sample data, to update the obstacle recognition algorithm.

4. The method according to claim 1 , wherein the displaying the to-be-labeled point cloud frame within the interface and displaying the camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface comprises:
saving a to-be-labeled point cloud frame in the point cloud data as a point cloud file; and
loading the point cloud file by using a point cloud visual tool, to display the point cloud frame in the point cloud visual tool.

5. The method according to claim 4 , wherein the displaying the to-be-labeled point cloud frame within the interface and displaying the camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface further comprises:
loading an image file corresponding to the camera image by using the point cloud visual tool, to display the camera image in the point cloud visual tool.

6. An apparatus for processing point cloud data, comprising:
at least one processor; and
a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
displaying a to-be-labeled point cloud frame within an interface;
displaying a camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface, the to-be-labeled point cloud frame being non-superimposed to the camera image;
determining, in response to an operation of selecting a displayed object in the point cloud frame by a user, an area encompassing the selected object in the point cloud frame;
projecting the area from the point cloud frame to the camera image, to obtain a projected area in the camera image;
displaying a mark in the projected area, for labeling, by the user, the selected object in the point cloud frame according to the mark indicating an object in the camera image; and
displaying simultaneously (i) the to-be-labeled point cloud frame, (ii) the camera image and (iii) label options for labeling the displayed object in the to-be-labeled point cloud frame within the interface, wherein the displayed to-be-labeled point cloud frame, the displayed camera image and the displayed label options are non-superimposed to each other.

7. The apparatus according to claim 6 , wherein the operations further comprise:
generating, in response to the labeling by the user on the selected object in the point cloud frame according to the mark, a labeled point cloud frame according to operation information of the labeling.

8. The apparatus according to claim 7 , wherein the operations further comprise:
training an obstacle recognition algorithm by using the labeled point cloud frame as sample data, to update the obstacle recognition algorithm.

9. The apparatus according to claim 6 , wherein the displaying the to-be-labeled point cloud frame within the interface and displaying the camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface comprises:
saving a to-be-labeled point cloud frame in the point cloud data as a point cloud file; and
loading the point cloud file by using a point cloud visual tool, to display the point cloud frame in the point cloud visual tool.

10. The apparatus according to claim 9 , wherein the displaying the to-be-labeled point cloud frame within the interface and displaying the camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface further comprises:
loading an image file corresponding to the camera image by using the point cloud visual tool, to display the camera image in the point cloud visual tool.

11. A non-transitory storage medium storing one or more programs, the one or more programs when executed by an apparatus, causing the apparatus to perform a method for processing point cloud data, comprising:
displaying a to-be-labeled point cloud frame within an interface;
displaying a camera image formed by photographing an identical scene at an identical moment as the point cloud frame within the interface, the to-be-labeled point cloud frame being non-superimposed to the camera image;
determining, in response to an operation of selecting a displayed object in the point cloud frame by a user, an area encompassing the selected object in the point cloud frame;
projecting the area from the point cloud frame to the camera image, to obtain a projected area in the camera image;
displaying-a mark in the projected area, for labeling, by the user, the selected object in the point cloud frame according to the mark indicating an object in the camera image; and
displaying simultaneously (i) the to-be-labeled point cloud frame, (ii) the camera image, and (iii) label options for labeling the displayed object in the to-be-labeled point cloud frame within the interface, wherein the displayed to-be-labeled point cloud frame, the displayed camera image and the displayed label options are non-superimposed to each other.

Assignees

• Beijing Baidu Netcom Sci & Tec

Inventors

• Feng Kaiwen
• Chen Zhuo
• Liu Bocong
• Liang Yibing
• Ma Yu
• Wang Haifeng

Classifications

• G06V10/774 Primary

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/41

  Interactive pattern learning with a human teacher · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G01S17/89

  for mapping or imaging · CPC title

• G01S17/86

  Combinations of lidar systems with systems other than lidar, radar or sonar, e.g. with direction finders · CPC title