Disclosed in the present invention is a head-mounted electronic vision aid device and an image magnification method thereof. The head-mounted electronic vision aid device comprising a memory unit, a processing unit, an image zooming unit, and at least one ranging unit; the ranging unit being configured to obtain distance data between a target object of interest to a user and the device and/or three-dimensional profile data of the object and output the data to the processing unit; the memory unit stores a correspondence table between the distance data and the magnification of the image zooming unit; the processing unit confirms the target object of interest to the user, performs operations on the distance data and/or the three-dimensional profile data of the object, and outputs an magnification matching the distance data to the image zooming unit according to the correspondence table; and the image zooming unit can automatically adjust to the matching magnification. For visually impaired users, accurate, intuitive and rapid automatic magnification of the target objects of interest can be realized on demand. Compared with the prior art, the repeated and tedious manual adjustment is avoided, and the user experience is greatly improved.

An augmented reality method using a mobile device having a display screen and a camera, the method includes receiving an image of a physical object from the camera, the physical object incorporating a machine-readable code, detecting the machine-readable code in the image of the physical object, decoding the machine-readable code, determining a website address of a website in response to decoding the machine-readable code, automatically navigating to the website according to the website address, and automatically receiving a video stream from the website and playing the video on the display screen of the mobile device superimposed over the image of the physical object.

A method of training a 3D structure detector to detect 3D structure in 3D structure representation, the method comprising the following steps: receiving, at a trainable 3D structure detector, a set of training inputs, each training input comprising at least one 3D structure representation; the 3D structure detector determining, for each training input, a set of predicted 3D objects for the at least one 3D structure representation of that training input; and training the 3D structure detector to optimize a cost function, wherein the cost function penalizes deviation from an expected geometric relationship between the set of predicted 3D objects determined for each training in put.

A computing system comprising one or more computing devices can obtain one or more images of an animal. The computing system can determine, using at least one of one or more machine-learned models, a plurality of joint positions associated with the animal based on the one or more images. The computing system can determine a body model for the animal. The computing system can estimate a body pose for the animal based on the one or more images, the plurality of joint positions, and the determined body model.

An image generation apparatus obtains a virtual viewpoint image generated based on captured images obtained by image capture of an object by a plurality of image capture devices from a plurality of viewpoints and three-dimensional shape data on the object, and removes noise in the virtual viewpoint image obtained, the noise being generated due to accuracy of the three-dimensional shape data.

A system and a method for object detection using augmented training dataset. The system includes a computing device, which is configured to: provide a two-dimensional (2D) image, extract feature vectors and estimate depths of 2D image pixels, generate a point cloud using the feature vectors and the depths, and project the point cloud using a new camera pose to obtain a projected image. The 2D image has a bounding box enclosing an object and labeled with the object. Each pixel within the bounding box is named bounding box pixel, each point in the point cloud corresponding to the bounding box pixel is named bounding box point, each image pixel in the projected image corresponding to the bounding box point is named projected bounding box pixel, and a projected bounding box is defined using the projected bounding box pixels and labeled with the object.

An example wear detection system receives first imaging data from one or more sensors associated with a work machine. The first imaging data comprises data related to at least one ground engaging tool (GET) of the work machine. The example system identifies a region of interest including data of the at least one GET within the first imaging data. Based on the identified region of interest, the example system controls a LiDAR sensor to capture second imaging data capturing the at least one GET that is of higher resolution than the first imaging data. The example system generates a three-dimensional point cloud of the at least one GET based on the second imaging data and determines a wear level or loss for the at least one GET based on the three-dimensional point cloud.

Removal of PII is provided. Sensor data is captured using sensors of a vehicle. Object detection is performed on the sensor data to create a sematic labeling of objects in the sensor data. A model is utilized to classify regions of the sensor data with a public or private labeling according to the sematic labeling and a PII filter corresponding to a jurisdiction of a current location of the vehicle. The sensor data is utilized in accordance with the public or private labeling.

Systems, methods, and devices described in the present disclosure provide technology for detecting features of interest depicted in a video stream supplied by a camera-equipped drone and adjusting the flight pattern of the drone to increase the amount of time in which the features of interest are within the line of sight of the drone. In addition, the present disclosure provides technology for detecting when events of interest occur at the features of interest and alerting personnel when the events occur. In some examples, the technology of the present disclosure also determines a format for an alert (e.g., a push notification) based on an event type so that personnel who receive the alert are apprised of the urgency of the event based on the format.

A method and an electronic apparatus for acquiring a floor map of a room layout are provided. The method includes acquiring a depth map and a two-dimensional (2D) image of an interior of a room; identifying a boundary line between a ceiling and a wall of the room in the acquired 2D image; determining valid pixels among pixels on the boundary line according to a first preset rule; identifying an electronic device in the room and determining a location of the electronic device; acquiring the floor map of the room layout by projecting three-dimensional (3D) coordinates of actual points corresponding to the valid pixels and 3D coordinates of the determined location of the electronic device onto a horizontal plane, wherein, the floor map of the room layout comprises connection lines between projections of the 3D coordinates of the actual points corresponding to the valid pixels and an icon indicating a type and the location of the electronic device, and wherein, the 3D coordinates of the actual points corresponding to the valid pixels are determined based on the depth map and the 2D image.