Abstract: A vehicular interior rearview mirror assembly includes a mirror head adjustably attached at a mounting base. The mirror head includes a mirror reflective element having a reflector-coated glass substrate having a front side and a rear side separated by a thickness of the glass substrate. A driver monitoring camera is accommodated by the mirror head and views through a camera region of the glass substrate of the mirror reflective element. A refraction-compensating element is disposed between a lens of the driver monitoring camera and the camera region of the glass substrate of the mirror reflective element. The refraction-compensating element has a first side and a second side that is not parallel to the first side. The refraction-compensating element offsets refraction of light that passes through the camera region of the glass substrate of the mirror reflective element.

Abstract: A medical image processing system includes an external device having a processor, and an endoscope system, in which the processor is configured to acquire a real-time medical video image from the endoscope system during endoscopy, receive a trigger signal having information regarding failure occurrence created based on a failure scene of the medical video image from the endoscope system, temporarily store the medical video image, extract a failure video image having the failure scene from the temporarily stored medical video image based on a reception timing of the trigger signal, and store the failure video image in a main storage region.

Abstract: A digital assistance for assisting a wearer of a hearing device to correctly arrange the hearing device with respect to an ear of the wearer, includes: a control configured to enable a preview of a first image to be displayed, wherein the first image is generated by a camera; wherein the control is also configured to enable a capturing of the first image, the first image showing at least the ear of the wearer with the hearing device arranged in and/or at the ear of the wearer; and wherein the digital assistance also comprises a comparator configured to perform a comparison between (1) the first image showing the at least the ear of the wearer with the hearing device and (2) a reference image.

Abstract: A network surveillance camera system includes a plurality of cameras configured to photograph a plurality of surveillance areas to acquire video information for the respective surveillance areas and transmit the acquired video information of each surveillance area to each channel, and a video receiving/search device connected to the plurality of cameras via a network and configured to receive the video information from the cameras in real time and search for the transmitted multi-channel video information, wherein the video receiving/search device includes a video search viewer configured to generate a thumbnail image and a metadata tag corresponding to an event of each channel based on video analysis information for each of the channels, display the generated thumbnail image and metadata tag, and search for the transmitted multi-channel video information.

Abstract: The disclosed embodiments relate to systems and methods, which among other things, acquire a digital rights management (DRM)-protected media content frame. The systems and methods package the DRM-protected media content frame to disguise the DRM-protected media content frame as a WebRTC-compliant video frame by including the DRM-protected media content frame as a payload for a WebRTC-compliant video frame container that includes a WebRTC-compliant coder and decoder (codec) header to produce a WebRTC-compliant video frame container package. The systems and methods may then transmit through a WebRTC video channel the WebRTC-compliant video frame container package.

Abstract: Certain aspects of the present disclosure provide techniques for performing 3D object detection. Such techniques may include obtaining one or more inputs associated with one or more two-dimensional (2D) views of a scene; selecting a set of 2D views of the scene from a plurality of 2D views of the scene based on the one or more inputs, the set of 2D views comprising a first 2D view of the scene and a second 2D view of the scene; and performing three-dimensional (3D) object detection in the scene based on the set of 2D views.

Abstract: A method and apparatus of utilizing an edge device comprising storing in a memory a pre-installed trained machine learning model, the machine learning model comprising a fixed base portion and a flexible portion. The method further comprises, when the edge device is installed, collecting data using one or more sensors on the edge device, and using the fixed base portion of the machine learning model on the data for a first aspect classification, while using the flexible portion of the machine learning model for a second aspect classification. The method further comprises calibrating the flexible portion of the machine learning model during the use.

Abstract: A system and method for processing a video stream captured during a clinical procedure. A method includes receiving image data from a scope during a procedure; forwarding the image data to a display as the scope is inserted into a patient; forwarding the image data to a cloud service to identify a landmark in the image data using a first machine learning model; receiving landmark details from the cloud service in response to a detected landmark and outputting the image data with landmark annotations; receiving and forwarding further image data to the display as the scope is further manipulated within the patient; forwarding the further image data to a cloud service that is configured to detect lesions using a second machine learning model; and receiving lesion details from the cloud service in response to a detected lesion and outputting the further image data with lesion annotations via the display.

Abstract: A system includes a first sensor system of a first modality and a second sensor system of a second modality. The system further includes a computing system that is configured to detect and identify objects represented in sensor signals output by the first and second sensor systems. The computing system employs a hierarchical arrangement of transformers to fuse features of first sensor data output by the first sensor system and second sensor data output by the second sensor system.

Abstract: A method for detecting a target pedestrian around a vehicle, a method for moving a vehicle, and a device are applied to an automobile electronic product. The method includes the following steps: monitoring a monitor area of a current vehicle in real time and acquiring a target trajectory; judging whether there is the target pedestrian in the monitor area according to the target trajectory, where if a judged result is true, there being the target pedestrian in the monitor area; and otherwise, performing no processing. According to the present application, the method for detecting a target pedestrian around a vehicle includes the following steps: judging whether there is a target vehicle owner around the vehicle; informing the vehicle owner; and then selecting, by the vehicle owner, whether a vehicle communication module is connected for communication. A first level communication initiative is mastered on an electronic rearview mirror, so that the problem that the vehicle owner is harassed is avoided.

Abstract: An endoscopy system according to the present disclosure includes an adapter for coupling a mobile device to an endoscope. A guide can be included that enables proper alignment of the adapter with the mobile device so that a generally centered image is presented on the mobile device from the endoscope. Further, various embodiments of the adapter can include integrated optical lenses.

Abstract: An invention disclosing a device and method for the counting and identification of bacterial, mold or yeast colonies based on hyperspectral image capture and processing. A device for acquisition of hyperspectral images comprising camera capable of hyperspectral imaging, a hole inspection lens array and a light source arranged in or around a light housing is described. A method comprising a deep learning algorithm is made available for the processing of acquired hyperspectral data in order to detect, identify and count bacterial, mold or yeast samples contained within a Petri dish. In particular, the use of a hole inspection lens array allows acquisition of hyperspectral images of the complete inner surface of a Petri dish which, combined with the use of a deep learning algorithm ensure a high level of accuracy and sampling frequency.

Abstract: Disclosed are systems and methods for selecting images using motion data. Video data and motion metadata can be received from a camera. A frame in the video data can be selected using the motion metadata. An image can be generated using the selected frame. A user interface comprising an element based on the image can be generated.

Abstract: Example apparatus disclosed herein includes a grouper to generate a first presence indication based on a threshold number of segments of image data including patterns of light values corresponding to a first human pulse. Disclosed example apparatus also include a face identifier to identify a first location associated with a human face in an environment associated with the image data based on thermal imaging data from a thermal imaging device, and output the first location associated with the human face in the environment associated with the image data.

Abstract: Systems and methods for a vision-based machine learning model for lane connectivity in autonomous or semi-autonomous driving. An example method includes obtaining images from a multitude of image sensors positioned about a vehicle; compute forward pass-through backbone networks of a machine learning model, wherein the output of the backbone networks are fused via a transformer network; aggregating information output from the transformer network across time and/or space; and determining lane connectivity information.

Abstract: A vehicle window color stripe processing method and a vehicle window color stripe processing apparatus, a storage medium and an electronic device are provided. The method includes: extracting a first vehicle window image in a to-be-processed image, where the first vehicle window image is an image with color stripes; and inputting the first vehicle window image into a pre-trained vehicle window color stripe processing model and obtaining a second vehicle window image output by the pretrained vehicle window color stripe processing model. The second vehicle window image is an image without color stripe, or, the intensity of color stripes in the second vehicle window image is lower than that of the color stripes in the first vehicle window image.

Abstract: A method for aquatic microplastic identification is provided. The method includes steps as follows: emitting a laser beam from a laser source, such that the laser beam passes through a liquid sample with MP samples in a sample channel and then a polarizer and is received by polarization camera; capturing a sample image of the MP samples by the polarization camera, wherein the sample image comprises interference patterns resulting from superposition of object and reference waves; and feeding the interference patterns into a morphology analyzing module for real-time tracking and analyzing by using a lightweight convolutional neural network (CNN) model, so as to classify the MP samples.

Abstract: Systems and techniques are provided for fusing raw sensor data captured by a camera sensor and one or more depth sensors to generate a dense depth map. An example process includes receiving raw camera data captured by a camera sensor and descriptive of a scene, receiving raw depth data captured by one or more depth-sensing sensors and descriptive of the scene, and providing the raw camera data and the raw depth data to a neural network, which is configured to fuse the raw camera data and the raw depth data. The example process can further include generating a depth map of the scene based on the fusion of the raw camera data and the raw depth data.

Abstract: Techniques for a computer-implemented service that utilizes a machine learning model to identify the salient and/or non-salient regions in a video frame are described. According to some embodiments, a computer-implemented method includes receiving a frame of a video at a content delivery service, generating, by a machine learning model of the content delivery service, a per pixel salience score map for the frame, and inserting, by the content delivery service, secondary content into the frame based at least in part on the per pixel salience score map for the frame.

Abstract: A data extraction device, which extracts an image for transmission from images obtained by capturing surroundings outside a vehicle with an imaging device mounted on the vehicle each time the vehicle travels an extraction distance, includes: processing circuitry configured to acquire an image obtained by capturing the surroundings outside the vehicle; determine, as an object to be analyzed, an image captured at a position advanced by the extraction distance from a position at which a previously extracted image has been captured, and calculate an obstruction distance which indicates an actual distance from an imaging start position of the image having been determined as the object to be analyzed to a position of an obstruction included in the determined image; output the calculated obstruction distance as a next extraction distance; and extract the image that has been determined as the object to be analyzed.