To detect a discrimination error in a type of an object. A calculation system includes a first device and a second device. The first device includes: a first object map generation unit configured to calculate, using first image information that is image information acquired by the first device, a first object map indicating a type of an object and a position of the object; and a first communication unit configured to transmit the first object map to the second device. The second device includes: a second object map generation unit configured to calculate, using second image information that is image information acquired by the second device, a second object map indicating a type of an object and a position of the object; and a comparison unit configured to compare the first object map and the second object map.

In an embodiment, a method comprises: receiving ambient sound; determining if the ambient sound includes a siren; in accordance with determining that the ambient sound includes a siren, determining a first location associated with the siren; receiving a camera image; determining if the camera image includes a flashing light; in accordance with determining that the camera image includes a flashing light, determining a second location associated with the flashing light; 3D data; determining if the 3D data includes an object; in accordance with determining that the 3D data includes an object, determining a third location associated with the object; determining a presence of an emergency vehicle based on the siren, detected flashing light and detected object; determining an estimated location of the emergency vehicle based on the first, second and third locations; and initiating an action related to the vehicle based on the determined presence and location.

A vehicular vision system includes a camera disposed at an in-cabin side of a windshield of a vehicle. Responsive to image processing at an ECU of captured frames of image data, the vehicular vision system detects an object present in a field of view of the camera. As the vehicle moves relative to the detected object, captured frames of image data are processed to determine a point of interest, present in multiple captured frames of image data, on the detected object. The vehicular vision system, via processing of captured frames of image data as the vehicle moves relative to the detected object, and based on the determined point of interest present in multiple captured frames of image data, estimates a location in three dimensional space of the determined point of interest and determines distance to the determined point of interest on the detected object.

An object orientation determination method includes generating pixel data on each of a plurality of unit pixels included in a region of interest of a point cloud acquired from an object, generating a plurality of candidate boxes using the generated pixel data, and determining, as a heading angle of an oriented bounding box, an inclination of a candidate box having a smallest cost among costs calculated on the plurality of candidate boxes. A cost of each of the plurality of candidate boxes is calculated based on positions of respective sides of a corresponding one of the plurality of candidate boxes and the pixel data.

A system includes an image sensor having a plurality of pixels that form a plurality of regions of interest (ROIs), and configured to operate at a frame rate higher than a threshold rate. The system also includes an image processing resource. The system further includes control circuitry configured to perform operations that include obtaining, from the image sensor, a full-resolution image of an environment. The full-resolution image contains each respective ROI of the plurality of ROIs. The operations also include selecting a particular ROI based on the full-resolution image, and detecting an object of interest in the particular ROI. The operations include determining a mode of operation by which subsequent image data generated by the particular ROI is to be processed. The operations further include processing, based on the mode of operation and the frame rate, the image data comprising a plurality of ROI images of the object of interest.

Systems and methods are provided for operating a vehicle, is provided. The method includes, by a vehicle control system of the vehicle, identifying map data for a present location of the vehicle using a location of the vehicle and pose and trajectory data for the vehicle, identifying a field of view of a camera of the vehicle, and analyzing the map data to identify an object that is expected to be in the field of view of the camera. The method further includes, based on (a) a class of the object, (b) characteristics of a region of interest in the field of view of the vehicle, or (c) both, selecting an automatic exposure (AE) setting for the camera. The method additionally includes causing the camera to use the AE setting when capturing images of the object, and using the camera, capturing the images of the object.

A method includes accessing a training sample including an image of a scene, depth measurements of the scene, and a predetermined 3D position of an object in the scene. The method includes training a 3D-detection model for detecting 3D positions of objects based the depth measurements and the predetermined 3D position, and training a 2D-detection model for detecting 2D positions of objects within images. Training the 2D-detection model includes generating an estimated 2D position of the object by processing the image using the 2D-detection model, determining a subset of the depth measurements that correspond to the object based on the estimated 2D position and a viewpoint from which the image is captured, generating an estimated 3D position of the object based on the subset of the depth measurements, and updating the 2D-detection model based on a comparison between the estimated 3D position and the predetermined 3D position.

A computer vision method and computer vision system can be used to process a time-based series of images. For a subject image of the time-based series, a light intensity value is identified for each pixel of a set of pixels of the subject image. A light intensity threshold is defined for the subject image based on a size of a bounding region for an object detected within a previous image of the time-based series captured before the subject image. A modified image is generated for the subject image by one or both of: reducing the light intensity value of each pixel of a lower intensity subset of pixels of the subject image that is less than the light intensity threshold, and increasing the light intensity value of each pixel of a higher intensity subset of pixels of the subject image that is greater than the light intensity threshold.

A method for tracking an at least partially occluded object. The method includes recognizing a non-occluded portion of the at least partially occluded object in an input image; generating a simulated image of the at least partially occluded object based on features of the non-occluded portion extracted from the input image; determining first coordinates of the at least partially occluded object in a first coordinate system; and converting the first coordinates of the at least partially occluded object in the first coordinate system into second coordinates in a second coordinate system defined in a display apparatus.

Systems and methods for estimating a height of an object from a monocular image are described herein. Objects are detected in the image, each object being indicated by a region of interest. The image is then cropped for each region of interest and the cropped image scaled to a predetermined size. The cropped and scaled image is then input into a convolutional neural network (CNN), the output of which is an estimated height for the object. The height may be represented by a mean of a probability distribution of possible sizes, a standard deviation, as well as a level of confidence. A location of the object may be determined based on the estimated height and region of interest. A ground truth dataset may be generated for training the CNN by simultaneously capturing a LIDAR sequence with a monocular image sequence.