A method for computing dimensions of an object in a scene includes: controlling, by a processor, a depth camera system to capture at least a frame of the scene, the frame including a color image and a depth image arranged in a plurality of pixels; detecting, by the processor, an object in the frame; determining, by the processor, a ground plane in the frame, the object resting on the ground plane; computing, by the processor, a rectangular outline bounding a projection of a plurality of pixels of the object onto the ground plane; computing, by the processor, a height of the object above the ground plane; and outputting, by the processor, computed dimensions of the object in accordance with a length and a width of the rectangular outline and the height.

A method for mapping an environment by an electronic device is described. The method includes obtaining a set of sensor measurements. The method also includes determining a set of voxel occupancy probability distributions respectively corresponding to a set of voxels based on the set of sensor measurements. Each of the voxel occupancy probability distributions represents a probability of occupancy of a voxel over a range of occupation densities. The range includes partial occupation densities.

Aspects of the present disclosure involve a system and a method for performing operations comprising: accessing a source image depicting a target structure; accessing one or more target images depicting at least a portion of the target structure; computing correspondence between a first set of pixels in the source image of a first portion of the target structure and a second set of pixels in the one or more target images of the first portion of the target structure, the correspondence being computed as a function of camera parameters that vary between the source image and the one or more target images; and generating a three-dimensional (3D) model of the target structure based on the correspondence between the first set of pixels in the source image and the second set of pixels in the one or more target images based on a joint optimization of target structure and camera parameters.

A method for coding information of a point cloud comprises obtaining the point cloud including a set of points in a three-dimensional space; partitioning the point cloud into a plurality of objects and generating occupancy information for each of the plurality of objects; and encoding the occupancy information by taking into account the distance between the plurality of objects.

Processes, systems, and devices for occlusion detection for video-based object tracking (VBOT) are described herein. Embodiments process video frames to compute histogram data and depth level data for the object to detect a subset of video frames for occlusion events and generate output data that identifies each video frame of the subset of video frames for the occlusion events. Threshold measurement values are used to attempt to reduce or eliminate false positives to increase processing efficiency.

A three-dimensional data creation method includes: creating first three-dimensional data from information detected by a sensor; receiving encoded three-dimensional data that is obtained by encoding second three-dimensional data; decoding the received encoded three-dimensional data to obtain the second three-dimensional data; and merging the first three-dimensional data with the second three-dimensional data to create third three-dimensional data.

Some implementations of the disclosure involve, at a device having one or more processors, one or more image sensors, and an illumination source, detecting a first attribute of an eye based on pixel differences associated with different wavelengths of light in a first image of the eye. These implementations next determine a first location associated with the first attribute in a three dimensional (3D) coordinate system based on depth information from a depth sensor. Various implementations detect a second attribute of the eye based on a glint resulting from light of the illumination source reflecting off a cornea of the eye. These implementations next determine a second location associated with the second attribute in the 3D coordinate system based on the depth information from the depth sensor, and determine a gaze direction in the 3D coordinate system based on the first location and the second location.

A method of automatically orienting symmetric and asymmetric food items, such as apples for example, is provided. Individual items of food are manipulated by a programmable manipulator within the view of one or more depth imaging cameras. Digital three dimensional characterizations of the surface of the food items are generated by the depth imaging camera or cameras and are utilized by a computer connected to the depth imaging camera or cameras to locate the stem and blossom of each food item. Asymmetric food items, such as apples with dropped shoulders as well as symmetric food items can be properly oriented and processed automatically.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for agent trajectory prediction using anchor trajectories.

Embodiments of the invention provide a camera array imaging architecture that computes depth maps for objects within a scene captured by the cameras, and use a near-field sub-array of cameras to compute depth to near-field objects and a far-field sub-array of cameras to compute depth to far-field objects. In particular, a baseline distance between cameras in the near-field subarray is less than a baseline distance between cameras in the far-field sub-array in order to increase the accuracy of the depth map. Some embodiments provide an illumination near-IR light source for use in computing depth maps.