Abstract: The present disclosure provides a target detection and model training method and apparatus, a device and a storage medium, and relates to the field of artificial intelligence, and in particular, to computer vision and deep learning technologies, which may be applied to smart city and intelligent transportation scenarios. The target detection method includes: performing feature extraction processing on an image to obtain image features of a plurality of stages of the image; performing position coding processing on the image to obtain a position code of the image; obtaining detection results of the plurality of stages of a target in the image based on the image features of the plurality of stages and the position code; and obtaining a target detection result based on the detection results of the plurality of stages.

Abstract: In various embodiments a computer-implemented method comprises generating, for a current video frame, a set of face motion data indicating whether a set of one or more faces detected in the current video frame has moved since a preceding video frame, generating a set of device motion data associated with one or more movements of an image capture device when capturing the current video frame, and generating a set of global motion data based on the set of face motion data and the set of device motion data, where the set of global motion data identifies a unique face motion when the set of face motion data indicates at least one face in the set of one or more faces has moved, and the set of device motion data indicates less than a threshold amount of motion of the image capture device when capturing the current video frame.

Abstract: A method for configuring a theme color of a terminal device includes obtaining a tone type of a to-be-switched-to picture when it is detected that a terminal device switches a screen picture, determining cool and warm color system information of the picture, identifying arrangement and proportion information of a representative color system of the picture, and configuring a theme color of the terminal device based on the tone type, the cool and warm color system information, and the arrangement and proportion information of the representative color system.

Abstract: The satellite image processing method includes: acquiring a first target satellite image; defogging the first target satellite image through a first neural network to acquire a first satellite image; and adjusting an image quality parameter of the first satellite image through a second neural network to acquire a second satellite image.

Abstract: A startup control method for a DC/DC converter is used for starting the DC/DC converter with energy transferred from a low-voltage side to a high-voltage side, and includes: determining a duty cycle or a frequency of a driving signal based on at least a voltage spike reference value and a voltage spike measurement value of a power switch at the low-voltage side; and outputting the driving signal to the power switch, thereby controlling the on and off of the power switch. The startup control method for a DC/DC converter of the disclosure can improve a reverse charging power of the converter to the maximum extent and reduce the time needed for charging a bus capacitor during a reverse startup process while ensuring voltage stress on the power switch not to exceed a limit.

Abstract: A method and device for detecting the key point of an image. The method includes acquiring the image, performing preprocessing to generate a multi-channel image with a specific type, and inputting the image into a detection model. The detection model includes a regression branch and an inter-frame residual branch. If the image type is the discrete frame, the system determines coordinates of multiple key points in the image based on a regression prediction value outputted by the regression branch. If the image type is continuous frame, the system determines coordinates of multiple key points in the image based on a regression prediction value outputted by the regression branch and a residual prediction value outputted by the inter-frame residual branch. The method can be used in the detection of discrete frame and discrete frame type images and can improve the stability of key point detection of images.

Abstract: A method for planning a path for a mobile object is disclosed. The method includes: generating a plurality of costmap layers based at least on locations of the obstacles in a navigation area, current locations and/or current planned paths of other mobile objects in the navigation area, and historical locations of other mobile objects in the navigation area; generating a master costmap based on the plurality of costmap layers; and planning a path to a target location based on the master costmap. An apparatus for planning a path for a mobile object is also disclosed. A controller for a mobile object, a mobile object and a computer-readable storage medium are also disclosed.

Abstract: A method of performing adjustment of a convergence speed for an imaging parameter is based on detecting a motion of a selected face in a sequence of image frames. When the detected motion meets predefined motion criteria, a motion vector corresponding to the characterized motion of the face is computed. A value for a convergence adjustment factor for adjusting a convergence speed of an imaging parameter of the camera is determined based on the computed motion vector. The convergence speed of the imaging parameter of the camera is adjusted based on the determined value of the convergence adjustment factor.

Abstract: Disclosed is a method for detecting densely occluded fish based on a YOLOv5 network, belonging to the technical field of fish images. The method includes a data set establishment and processing part, a model training part and a model testing part; the data set establishment and processing part includes data collection of fish pictures, data labelling and data division of the fish pictures; and the data division is to divide data into a training set, a verification set and a test set.

Abstract: Example synchronization signal sending and receiving methods and apparatus are described. In one example method, a terminal device determines a target frequency resource. A frequency domain position of the target frequency resource is determined based on a frequency domain position offset and a frequency interval of synchronization channels. The terminal device receives a synchronization signal by using the target frequency resource.

Abstract: Embodiments of the present disclosure provide a method and apparatus for detecting a target object in an image. The method includes: performing following prediction operations using a pre-trained neural network: detecting a target object in a two-dimensional image to determine a two-dimensional bounding box of the target object; and determining a relative position constraint relationship between the two-dimensional bounding box of the target object and a three-dimensional projection bounding box obtained by projecting a three-dimensional bounding box of the target object into the two-dimensional image; and the method further including: determining the three-dimensional projection bounding box of the target object, based on the two-dimensional bounding box of the target object and the relative position constraint relationship between the two-dimensional bounding box of the target object and the three-dimensional projection bounding box.

Abstract: In a communication method, a terminal device obtains a code division multiplexing capability of a physical resource of an uplink control channel, where the code division multiplexing capability is a length of orthogonal cover codes for multiplexing terminal devices in a code division manner on a unit resource in the physical resource of the uplink control channel, and the code division multiplexing capability configured by higher layer signaling. The terminal device further determines the physical resource based on the code division multiplexing capability, and sends uplink control information on the physical resource.

Abstract: Devices, methods, and non-transitory program storage devices (NPSDs) for an improved, so-called “hybrid” image registration process are disclosed herein, comprising: obtaining a first set of captured images, wherein the first set of captured images comprises a reference image and one or more bracketed images; and for each of the one or more bracketed images: performing a first (e.g., global) registration operation and a second (e.g., dense, or other localized) registration operation on the bracketed image with respect to the reference image, wherein each of the first and second registration operations produces an output; generating a blend map for the bracketed image, wherein each value in the blend map indicates whether to use the first or second registration operation output for a corresponding one or more pixels when registering the bracketed image with the reference image; and registering the bracketed image with the reference image, according to the generated blend map.

Abstract: Aspects disclosed herein include fully or partially degradable polymers, liquid formulations comprising such polymers, methods for treating or managing a condition in a subject using such polymers, methods for using such polymers, and methods of synthesizing such polymers. Aspects disclosed herein also include monomers suitable for forming a partially or fully degradable polymer and methods for forming such monomers.

Abstract: A method for training a point cloud processing model and a method for performing instance segmentation on a point cloud are provided. The method includes: labeling an unlabeled point cloud according to a labeled point cloud to obtain a sample point cloud; inputting the sample point cloud to a point cloud processing model to obtain first predicted semantic information of the sample point cloud and a first predicted offset of the sample point cloud; determining a training loss according to the first predicted semantic information, the first predicted offset, a sample label corresponding to the sample point cloud, and original coordinate information of the sample point cloud; and training the point cloud processing model with the training loss.

Abstract: Provided are a joint perception model training method, a joint perception method, a device, and a storage medium. The joint perception model training method includes: acquiring sample images and perception tags of the sample images; acquiring a preset joint perception model, where the joint perception model includes a feature extraction network and a joint perception network; performing feature extraction on the sample images through the feature extraction network to obtain target sample features; performing joint perception through the joint perception network according to the target sample features to obtain perception prediction results; and training the preset joint perception model according to the perception prediction results and the perception tags, where the joint perception includes executing at least two perception tasks.

Abstract: A method of generating a virtual idol is provided. The method includes: during obtaining of a virtual idol, feature information of a target object may be obtained, and then the feature information of the target object is used as a generation basis for the virtual idol, so that a target virtual face matching with the feature information may be determined from a preset face material library in a targeted manner, and a target motion video matching with the feature information may be determined from a preset motion video library; and then the virtual face and a facial image in the motion video are fused, which may generate the virtual idol for a scenario where the target object is endorsed, so that the virtual idol may subsequently be used to endorse the target object.

Abstract: A sequence-based signal processing method and apparatus are provided. A sequence meeting a requirement for sending a signal by using a physical uplink control channel (PUCCH) is determined. The sequence is a sequence {fn} consisting of 12 elements, fn represents an element in the sequence {fn}, and the determined sequence {fn} is a sequence meeting a preset condition. Then, the 12 elements in the sequence {fn} are respectively mapped to 12 subcarriers, to generate a first signal, and the first signal is sent. By using the determined sequence, when the signal is sent by using the PUCCH, a low correlation between sequences can be maintained, and a relatively small peak-to-average power ratio (PAPR) value and a relatively small cubic metric (CM) value can be maintained. Therefore, a requirement of a communication application environment in which the signal is sent by using the PUCCH is met.

Abstract: The present disclosure provides a three-level rectification DC/DC converter including primary and secondary circuits and a resonant tank circuit. A voltage between two primary terminals is a first voltage. The secondary circuit includes a switch bridge arm and a capacitor bridge arm. The switch bridge arm includes four switches serially connected. A node between the first and second switches is connected to the first secondary terminal, a node between the third and fourth switches is connected to the second secondary terminal, and a node between the second and third switches is connected between two capacitors of the capacitor bridge arm. In two consecutive periods of the first voltage, the first and fourth switches are in an on state for a preset time length after two falling edges respectively, and the second and third switches are in the on state for the preset time length after two rising edges respectively.