Provided are a method and an apparatus for reporting and handling an exception in computing power sharing, a storage medium, and a terminal device. The method for reporting an exception in computing power sharing includes: detecting a current hardware state and a current battery state; and reporting an exception to a network unit, in a case that the hardware state or the battery state reaches a preset exception threshold, or in a case that a change of the hardware state or a change of the battery state reaches a preset reporting threshold. The method for handling an exception in computing power sharing includes: receiving an exception reported from a cooperative computing terminal; determining a total workload assigned to the cooperative computing terminal and a remaining workload of the cooperative computing terminal; and determining, based on the exception and the remaining workload, to reassign the remaining workload or the total workload.

This application discloses a neural network model training method, an image processing method, and an apparatus in the field of artificial intelligence. The method includes: inputting training data to a neural network model for feature extraction, and obtaining a first weight gradient of the neural network model based on an extracted feature; obtaining a candidate weight parameter, where a partial derivative of a function value of a target loss function to the candidate weight parameter is 0, the function value of the target loss function is determined based on a function value of a second loss function corresponding to a first prediction label, and the function value of the second loss function corresponding to the first prediction label indicates a difference between the candidate weight parameter and a weight parameter of the neural network model and a difference between a weight variation and the first weight gradient.

The present disclosure relates to an image analysis method, system, and computer program. The image analysis method of the present disclosure includes: receiving a query image; extracting one or more regions of interest from the query image; calculating a first feature for each of the regions of interest by respectively applying the regions of interest to one or more ROI (region of interest) feature extraction models independently learned in order to extract features of the regions of interest; and calculating analysis values of the query image by applying the first features of the regions of interest to a pre-learned integration analysis model. According to the present disclosure, it is possible to reduce the influence on an analysis model by an error that training data created for map learning of an entire image may have, and it is also possible to increase learning accuracy and objectivity of a deep neural network.

The present disclosure provides a character recognition model training method and apparatus, a character recognition method and apparatus, a device and a medium, relating to the technical field of artificial intelligence, and specifically to the technical fields of deep learning, image processing and computer vision, which can be applied to scenarios such as character detection and recognition technology. The specific implementing solution is: partitioning an untagged training sample into at least two sub-sample images; dividing the at least two sub-sample images into a first training set and a second training set; where the first training set includes a first sub-sample image with a visible attribute, and the second training set includes a second sub-sample image with an invisible attribute; performing self-supervised training on a to-be-trained encoder by taking the second training set as a tag of the first training set, to obtain a target encoder.

An electronic apparatus includes a memory for storing a classification network including a plurality of feature extraction layers. The electronic apparatus also includes a processor for acquiring a class score corresponding to an object, which is output from the classification network, by inputting a training image including the object to the classification network, acquiring a final loss value, based on a plurality of activation maps respectively output from the plurality of feature extraction layers and the class score, and controlling the classification network, based on the final loss value.

Methods for determining a value of an intrinsic geometrical parameter of a geometrical feature characterizing a physical object, and for classifying a scene into at least one geometrical shape, each geometrical shape modeling a luminous object. A singular light distribution characterized by a first wavelength and a position of singularity is projected onto the physical object. Light excited by the singular light distribution that has interacted with the geometrical feature and that impinges upon a detector is detected and a return energy distribution is identified and quantified at one or more positions. A deep learning or neural network layer may be employed, using the detected light as direct input of the neural network layer, adapted to classify the scene, as a plurality of shapes, static or dynamic, the shapes being part of a set of shapes predetermined or acquired by learning.

The present disclosure relates to a cargo protection method, device and system, and a non-transitory computer-readable storage medium, relating to the technical field of unmanned aerial vehicles. The method of the present disclosure includes: determining whether an unmanned aerial vehicle is in a falling state or not according to a current acceleration in a vertical direction of the unmanned aerial vehicle and a current vertical distance from the unmanned aerial vehicle to the ground; and opening at least one airbag in a cargo hold of the unmanned aerial vehicle in a case where the unmanned aerial vehicle is in the falling state to protect a cargo in the cargo hold.

An electronic apparatus and a controlling method thereof are provided. The electronic apparatus providing augmented reality (AR) content includes a display, a camera and a processor configured to display augmented reality (AR) content through the display, detect a hand of a user from image obtained through the camera, and identify a first interaction of the hand with the AR content based on a size of the hand, wherein the size of the hand is obtained based on the information about an object provided through the display.

A method for configuring an object detection system includes providing annotated training data comprising image data with defined assignments to at least one object, and training a neural network with a first neural sub-network, which is provided to compress the image data. The first neural sub-network is connected to at least one further neural sub-network. The at least one further neural sub-network is configured to detect an object from the compressed training data. The first neural sub-network is parameterized in such a manner that the object is detected using the at least one further sub-network in a defined quality. The neural sub-networks are trained jointly.

A method for calculating an anchoring area of a ship includes steps of obtaining ship trajectories of ships within a certain time period in an anchorage; screening out ship trajectories including an anchoring process and eliminating trajectory points in a non-anchored state in the ship trajectories to obtain anchoring trajectories of anchored ships; clustering anchoring points in each of the anchoring trajectories, using a cluster center as an anchoring position point of each of the anchored ships; establishing an anchoring data set according to the anchoring position points; selecting anchoring data records in a predetermined time period in the anchoring data set; establishing an anchored ship position point set corresponding to the predetermined time period; and establishing Thiessen polygons corresponding to the anchoring position points; calculating an area of each of the Thiessen polygons to obtain an anchoring area of a corresponding anchored ship.