Speed estimation systems and methods without camera calibration

Patent metadata

Field	Value
Publication number	US-2025285298-A1
Application number	US-202519218834-A
Country	US
Kind code	A1
Filing date	May 27, 2025
Priority date	Jan 29, 2021
Publication date	Sep 11, 2025
Grant date	—

Abstract

A speed estimation system includes: a detection module having a neural network configured to: receive a time series of images, the images including a surface having a local geometry; detect an object in the time series of images on the surface; determine pixel coordinates of the object in the time series of images, respectively; determine bounding boxes around the object in the time series of images, respectively; determine local mappings, which are not a function of global parameters describing the local geometry of the surface, between pixel coordinates and distance coordinates for the time series of images based on the bounding boxes around the object in the time series of images, respectively; and a speed module configured to determine a speed of the object traveling relative to the surface based on the distance coordinates determined for the time series of images.

First claim

1 . A speed estimation system comprising:
a detection module having a neural network configured to:
receive a time series of images, the images including a surface having a local geometry;
detect an object in the time series of images on the surface;
determine pixel coordinates of the object in the time series of images, respectively;
determine two dimensional (2D) bounding boxes around the object in the time series of images, respectively;
determine local mappings, which are not a function of global parameters describing the local geometry of the surface, between pixel coordinates and distance coordinates for the time series of images based on the 2D bounding boxes around the object in the time series of images, respectively; and
a speed module configured to determine a speed of the object traveling relative to the surface based on the distance coordinates determined for the time series of images; wherein the detection module is configured to determine the local mappings using Jacobians.

2 . The speed estimation system of claim 1 further comprising an averaging module configured to determine an average speed of the object based on an average of multiple instances of speed of the object in the time series of images.

3 . The speed estimation system of claim 2 wherein the averaging module performs median filtering on the speeds of the object in the time series of images before determining the average speed.

4 . The speed estimation system of claim 1 wherein the object on the surface is a vehicle on a road and the surface is planar.

5 . The speed estimation system of claim 1 further comprising a tracking module configured to generate a track for movement of the object based on the pixel coordinates of the images, respectively.

6 . The speed estimation system of claim 5 wherein the tracking module is configured to track the object in the images using the simple online and realtime tracking (SORT) tracking algorithm.

7 . The speed estimation system of claim 5 wherein the tracking module is configured to disable the determination of the speed of the object when a number of detections of the object in the images is less than a predetermined number.

8 . The speed estimation system of claim 5 wherein the tracking module is configured to disable the determination of the speed of the object when the object is not moving.

9 . The speed estimation system of claim 1 wherein the detection module includes:
a feature detection module configured to detect features in one of the time series of images;
a regional proposal module configured to, based on the features of the one of the time series of images, propose a region of the one of the images within which the object is present;
a regional pooling module configured to pool features within the region to create pooled features;
a classifier module configured to determine the classification of the object based on the pooled features; and
a bounding module configured to determine the 2D bounding box for the one of the images based on the pooled features.

10 . The speed estimation system of claim 1 wherein the detection module includes a convolutional neural network.

11 . The speed estimation system of claim 10 wherein the convolutional neural network of the detection module executes the Faster-regions with convolutional neural network (Faster-RCNN) object detection algorithm.

12 . The speed estimation system of claim 1 wherein:
the neural network of the detection module is further configured to:
detect a second object in the time series of images on the surface;
determine second pixel coordinates of the second object in the time series of images, respectively;
determine second 2D bounding boxes around the second object in the time series of images, respectively;
determine second local mappings, which are not a function of global parameters describing the local geometry of the surface, between pixel coordinates and distance coordinates for the time series of images based on the second 2D bounding boxes around the second object in the time series of images, respectively; and

the speed module is configured to determine a second speed of the second object traveling relative to the surface based on the second distance coordinates determined for the time series of images,
wherein the detection module is configured to determine the second local mappings using second Jacobians.

13 . The speed estimation system of claim 12 further comprising an average speed module configured to determine an average speed based on an average of the speed and the second speed.

14 . The speed estimation system of claim 1 wherein the detection module is configured to receive the time series of images from a monocular camera.

15 . The speed estimation system of claim 14 wherein the monocular camera is a pan, tilt, zoom (PTZ) camera.

16 . The speed estimation system of claim 1 wherein the speed module is configured to determine the speed of the object further based on a change in the pixel coordinates from a first one of the images to a second one of the images.

17 . The speed estimation system of claim 1 wherein the neural network is trained to determine the local mappings between pixel coordinates and distance coordinates using Jacobians.

18 . (canceled)

19 . The speed estimation system of claim 17 wherein the neural network of the detection module is trained to determine a Jacobians and the inverse of the Jacobian from a 3D bounding boxes of an object.

20 . The speed estimation system of claim 17 wherein the neural network of the detection module is trained to determine the Jacobians further based on a length of the object and a width of the object.

21 . The speed estimation system of claim 1 wherein the detection module is configured to receive the time series of images from a video source via a network.

22 . The speed estimation system of claim 1 wherein the speed module is configured to determine the speed of the object without stored calibration parameters of a camera.

23 . A routing system, comprising:
the speed estimation system of claim 1 ; and a route module configured to:
determine a route for one of a mobile device and a vehicle based on the speed of the object; and
transmit the route to the one of the mobile device and the vehicle.

24 . A signaling system, comprising:
the speed estimation system of claim 1 ; and a signal control module configured to:
determine a timing for a traffic signal based on the speed of the object; and
control timing of the traffic signal based on the timing.

25 . A method for estimating a speed of an object in a time series of images using a neural network, comprising:
receiving the time series of images, the images including a surface having a local geometry; by the neural network:
detecting an object in the time series of images on the surface;
determining pixel coordinates of the object in the time series of images, respectively;
determining two dimensional (2D) bounding boxes around the object in the time series of images, respectively;
determining local mappings, which are not a function of global parameters describing the local geometry of the surface, between pixel coordinates and distance coordinates for the time series of images based on the 2 D_bounding boxes around the object in the time series of images, respectively; and
determining a speed of the object tr

Assignees

• Naver Corp

Inventors

• Revaud Jérome
• Cabon Yohann
• Morat Julien

Classifications

• G06V20/64

  Three-dimensional [3D] objects · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/30252

  Vehicle exterior; Vicinity of vehicle · CPC title

• G06T7/74

  involving reference images or patches · CPC title