Disclosed are a method for processing an image, an electronic device, and a storage medium. The method includes: acquiring a plurality of point cloud grids from a gridded point cloud map; determining each reference plane with the number of sampling points matched being greater than or equal to a threshold from the initial planes corresponding respectively to the plurality of the point cloud grids; correcting the initial planes corresponding respectively to the plurality of point cloud grids based on each reference plane to obtain target planes corresponding respectively to the plurality of point cloud grids; and performing denoise processing on sampling points based on the target plane of each point cloud grid.

A multi-camera vehicle vision system and method. In one embodiment a map is generated about a moving vehicle. Frames of image data are provided with a series of cameras extending along a surface of the vehicle. The image data frames are processed to identify an object of interest. An object of interest is classified among a set of object types and location of an identified object of interest is determined. Object type and location information is provided to a control unit spaced apart from the cameras via a data link. Road map data is generated to illustrate changes in position of the moving vehicle along a roadway based on data other than the image data provided by the cameras. A display of the road map data is generated with the object type and location information overlaid on the road map data to indicate object location relative to the vehicle.

A data acquisition system of a vehicle includes an image capture device, a communication interface, and a controller communicatively coupled to the image capture device and communicatively coupled to the communication interface. Processors of the controller are configured to calibrate an image-distance relationship value of an identified component of a first image captured by the image capture device corresponding to a known feature based on established metrics of the known feature. The processors are also configured to provide control of the vehicle or activation of an alert system of the vehicle via the communication interface based on the image-distance relationship value.

The present disclosure provides a multiple target tracking method and apparatus, a calculating device and a storage medium, so as to solve the problem of inaccurate multiple target tracking in the prior art. The target tracking method comprises: obtaining a prediction box of a target in a current frame according to tracklets of one or more targets in historical frames; performing target detection on the current frame to obtain one or more detection boxes, wherein the detection boxes comprise a high-quality box and a medium-quality box; matching each prediction box with the detection box according to the similarity of the prediction box and the detection box; and in response to a prediction box being unmatched with the high-quality box but matched with the medium-quality box, determining that the target is in a tracking state in the current frame.

Systems and methods for determining the drivable space of a road, for applications such as autonomous navigation. To determine the non-drivable space under another vehicle, systems and methods of embodiments of the disclosure generate 3D bounding boxes from 2D bounding boxes of objects in captured roadway images, and from various geometric constraints. Image portions may be labeled as drivable or non-drivable according to projections of these 3D bounding boxes onto their road surfaces. These labeled images, along with accompanying semantic information, may be compiled to form training datasets for a machine learning model such as a CNN. The training datasets may train the CNN to classify input image portions into drivable and non-drivable space, for applications such as autonomous navigation.

In various examples, systems and methods are described that generate scene flow in 3D space through simplifying the 3D LiDAR data to “2.5D” optical flow space (e.g., x, y, and depth flow). For example, LiDAR range images may be used to generate 2.5D representations of depth flow information between frames of LiDAR data, and two or more range images may be compared to generate depth flow information, and messages may be passed—e.g., using a belief propagation algorithm—to update pixel values in the 2.5D representation. The resulting images may then be used to generate 2.5D motion vectors, and the 2.5D motion vectors may be converted back to 3D space to generate a 3D scene flow representation of an environment around an autonomous machine.

An automated driving function in a motor vehicle comprises: a processor circuit of the motor vehicle recognizes respective individual images of an environment of the motor vehicle from sensor data of a least one sensor of the motor vehicle by means of at least one object classifier. At least one relational classifier using the object data for at least some of the individual objects additionally recognizes a respective pairwise object relation with the aid of predetermined relation features of the individual objects in the respective individual image determined from the sensor data, which relation is described by relational data, and an aggregation module is used to aggregate the relational data throughout multiple consecutive individual images to produce aggregation data, which describe aggregated object relations.

Disclosed herein is a device for filtering a point cloud. The device may include processor configured to receive a plurality of sensed points representing distance measurements to points in an area around an entity. The processor may also be configured to determine, for each sensed point of the plurality of sensed points, one or more probabilities that the sensed point is associated with an obstacle in the area around the entity, wherein each probability of the one or more probabilities is based on a corresponding filter of one or more filters. The processor may also be configured to generate a filtered point cloud from the plurality of sensed points based on, for each sensed point, a hierarchical combination of the one or more probabilities.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating motion detection based on optical flow. One of the methods includes obtaining a first image of a scene in an environment taken by an agent at a first time point and a second image of the scene at a second later time point. A point cloud characterizing the scene in the environment is obtained. A predicted optical flow is determined between the first image and the second image. A respective initial flow prediction for the point that represents motion of the point between the two time points is determined. A respective ego motion flow estimate for the point that represents a motion of the point induced by ego motion of the agent is determined. A respective motion prediction that indicates whether the point was static or in motion between the two time points is determined.

Techniques are discussed herein for generating three-dimensional (3D) representations of an environment based on two-dimensional (2D) image data, and using the 3D representations to perform 3D object detection and other 3D analyses of the environment. 2D image data may be received, along with depth estimation data associated with the 2D image data. Using the 2D image data and associated depth data, an image-based object detector may generate 3D representations, including point clouds and/or 3D pixel grids, for the 2D image or particular regions of interest. In some examples, a 3D point cloud may be generated by projecting pixels from the 2D image into 3D space followed by a trained 3D convolutional neural network (CNN) performing object detection. Additionally or alternatively, a top-down view of a 3D pixel grid representation may be used to perform object detection using 2D convolutions.