Untrained systems and methods for vehicle speed estimation

Patent metadata

Field	Value
Publication number	US-2023032420-A1
Application number	US-202117365439-A
Country	US
Kind code	A1
Filing date	Jul 1, 2021
Priority date	Jul 1, 2021
Publication date	Feb 2, 2023
Grant date	—

Abstract

A speed estimation system includes: a detection module configured to determine bounding boxes of an object moving on a surface in images, respectively, captured using a camera; a solver module configured to, based on the bounding boxes, determine a homography of the surface by solving an optimization problem, where the solver module is not trained; and a speed module configured to, using the homography, determine a speed that the object is moving on the surface.

First claim

What is claimed is:

1 . A speed estimation system, comprising:
a detection module configured to determine bounding boxes of an object moving on a surface in images, respectively, captured using a camera; a solver module configured to, based on the bounding boxes, determine a homography of the surface by solving an optimization problem, wherein the solver module is not trained; and a speed module configured to, using the homography, determine a speed that the object is moving on the surface.

2 . The speed estimation system of claim 1 wherein the solver module configured to determine the homography using an optimization algorithm.

3 . The speed estimation system of claim 2 wherein the optimization algorithm is configured based on an assumption that the surface is planar.

4 . The speed estimation system of claim 2 wherein the optimization algorithm is configured based on an assumption that pixels of the images are square.

5 . The speed estimation system of claim 2 wherein the optimization algorithm is configured based on an assumption that images captured by the camera do not have horizontal skew and do not have vertical skew.

6 . The speed estimation system of claim 2 wherein the optimization algorithm is configured based on an assumption that a principal point is at a center of the image.

7 . The speed estimation system of claim 2 wherein the optimization algorithm is configured based on an assumption of zero roll of the camera.

8 . The speed estimation system of claim 1 wherein the solver module includes a plurality of encoder modules configured to determine the homography.

9 . The speed estimation system of claim 8 wherein the encoder modules have the Transformer architecture.

10 . The speed estimation system of claim 1 wherein the solver module is configured to filter the homography before the homography is used to determine the speed that the object is moving on the surface.

11 . The speed estimation system of claim 1 wherein the object is a vehicle.

12 . A speed estimation method, comprising:
determining bounding boxes of an object moving on a surface in images, respectively, captured using a camera; without training, based on the bounding boxes, determining a homography of the surface by solving an optimization problem; and using the homography, determine a speed that the object is moving on the surface.

13 . The speed estimation method of claim 12 wherein determining the homography includes determining the homography using an optimization algorithm.

14 . The speed estimation method of claim 13 wherein the optimization algorithm is configured based on an assumption that the surface is planar.

15 . The speed estimation method of claim 13 wherein the optimization algorithm is configured based on an assumption that pixels of the images are square.

16 . The speed estimation method of claim 13 wherein the optimization algorithm is configured based on an assumption that images captured by the camera do not have horizontal skew and do not have vertical skew.

17 . The speed estimation method of claim 13 wherein the optimization algorithm is configured based on an assumption that a principal point is at a center of the image.

18 . The speed estimation method of claim 13 wherein the optimization algorithm is configured based on an assumption of zero roll of the camera.

19 . The speed estimation method of claim 12 wherein determining the homography includes determining the homography using a plurality of encoder modules configured to determine the homography.

20 . The speed estimation method of claim 19 wherein the encoder modules have the Transformer architecture.

21 . The speed estimation method of claim 12 further comprising filtering the homography before the homography is used to determine the speed that the object is moving on the surface.

22 . A speed estimation system, comprising:
a means for determining bounding boxes of an object moving on a surface in images, respectively, captured using a camera; an untrained means for, based on the bounding boxes, determining a homography of the surface by solving an optimization problem; and a means for, using the homography, determining a speed that the object is moving on the surface.

Assignees

• Naver Corp

Inventors

• Revaud Jérome
• Cabon Yohann

Classifications

• G08G1/0116

  from roadside infrastructure, e.g. beacons · CPC title

• G08G1/0133

  for classifying traffic situation · CPC title

• G08G1/052

  with provision for determining speed or overspeed {(speed measuring in general G01P)} · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/30236

  Traffic on road, railway or crossing · CPC title