Jingdong Technology Information Technology Co., Ltd.

A quick connection device includes: a sleeved connected with a first member and a moving assembly movably disposed in the sleeve. The moving assembly includes: a support base disposed in the sleeve; a mounting base, configured to be connected with a second member and movable along the axial direction relative to the support base; a locking member, fixed to the mounting base and configured to cooperate with the support base to achieve a locking state and an unlocking state and restrict movement of the mounting base towards a direction away from the support base when in the locking state and allow movement of the mounting base towards a direction away from the support base when in the unlocking state; and a limiting member, configured to restrict movement of the moving assembly in the sleeve when the mounting base is in an initial position corresponding to the locking state.

A service request processing method includes that: a target domain name resolution request corresponding to a target federated service to be accessed by a first deployment unit in a current working cluster is generated; the target domain name resolution request is sent to a federated domain name cluster, and the federated domain name cluster processes the target domain name resolution request to determine a target IP address; an IP address of a second deployment unit for implementing the target federated service in a target working cluster is determined based on the target IP address and based on the federated gateway address and the federated endpoint set that correspond to each federated service determined by the federated management cluster; and the target federated service request is sent to the second deployment unit based on the IP address of the second deployment unit

A method for recognizing a text, and an apparatus. A specific embodiment of the method comprises: obtaining feature maps, the feature maps being obtained by means of performing text instance segmentation on an image having a text to be recognized; constructing a relationship graph according to the feature maps, wherein in the relationship graph, each node represents a pixel in a feature map, and each edge represents that a similarity measure of spatial semantic features of two connected nodes is greater than a target threshold, and a spatial semantic feature of a node comprises a type feature and a position feature of a pixel represented by the node; utilizing a pre-trained graph convolutional network to perform processing on the relationship graph, and obtaining a first text feature corresponding to the image; and generating a text recognition result for the image according to the first text feature.

The disclosure provides a network card packet receiving processing method and apparatus. The method comprises: parsing received message data, and determining pre-fetching control information of a receiving queue corresponding to the message data; determining corresponding pre-fetched memory block scheduling information based on the pre-fetching control information; and performing a memory block pre-fetching scheduling operation on the receiving queue in a network card based on the pre-fetched memory block scheduling information, so as to generate a pre-fetching request to acquire corresponding memory block information. According to the network card packet receiving processing method provided in the disclosure, corresponding pre-fetched memory block scheduling information is generated by receiving pre-fetching control information corresponding to message data, and packet receiving scheduling among a plurality of receiving queues can be achieved based on the pre-fetched memory block scheduling information, which improves the pre-fetched memory block scheduling efficiency of a network card, thereby improving the performance of the whole network card in receiving the message data.

A robot relocalization method and apparatus, and a storage medium and an electronic device. The method comprises: performing feature coding on local visual features of image data to obtain image vector features, and searching, in a pre-set key frame database, for the image vector features subjected to dimension reduction processing, so as to obtain a plurality of search results (S 120 ); solving the pose of a robot according to each search result, so as to obtain a first pose initial value, and performing matching on laser point cloud data in a pre-set two-dimensional grid map according to speed data and displacement data, so as to obtain a first matching result (S 130 ); and generating a second matching result according to the first pose initial value and the first matching result, and performing robot relocalization according to a second pose initial value, the first matching result and the second matching result.

A method for controlling a robotic arm includes: acquiring a first depth image and a first color image of a target object; predicting first predicted values and second predicted values of respective candidate actions of the robotic arm based on the first depth image and the first color image, in which the first predicted value represents a probability of the robotic arm separating the target object from an adjacent object by executing the corresponding candidate action; and the second predicted value represents a probability of the robotic arm successfully grabbing the target object by executing the corresponding candidate action; determining a target action based on the first predicted values and the second predicted values of the candidate actions; and controlling the robotic arm to execute the target action.

A container scheduling method, includes: from a queue to be scheduled, obtaining N pods to be scheduled; recording a first version number of each schedulable node in a cluster; traversing second attribute information of each node on the basis of first attribute information of each pod to be scheduled, so as to determine a candidate node corresponding to each pod to be scheduled, and a second version number corresponding to the candidate node; and for any pod among the pods to be scheduled, if a second version number of a candidate node corresponding to the pod matches with the recorded first version number, using the candidate node to schedule the pod.

A training method, apparatus and system for a feature extraction network of a 3D mesh model are provided. The method includes dividing a training 3D mesh model into a plurality of patches which do not overlap with each other, dividing the plurality of patches into first-type patches and second-type patches, and using mask embedding as a feature encoding of each second-type patch; inputting geometric representation information and positional representation information of each first-type patch into a feature extraction network; determining predicted geometric representation information of each face based on a feature encoding of each first-type patch output from the feature extraction network, the mask embedding, and positional representation information of each second-type patch, and adjusting parameters of the feature extraction network based on differences between the predicted geometric representation information and geometric representation information of the each face.

A method for training a speech recognition model, includes: constructing an initial speech recognition model including a first network having a first initial parameter and a second network having a second initial parameter; fixing the second initial parameter, calculating a contrastive learning loss function, and performing self-supervised training on the first network according to the contrastive learning loss function to adjust the first initial parameter to a first intermediate parameter; fixing the first intermediate parameter, calculating a first joint loss function, and performing training on the second network according to the first joint loss function to adjust the second initial parameter to a second intermediate parameter; and calculating a second joint loss function, and performing training an the first network and the second network according to the second joint loss function to adjust the first intermediate parameter and the second intermediate parameter to obtain a target speech recognition model.

The present disclosure relates to a track travelling device, comprising a vehicle body, a connection device and two groups of wheels. The two groups of wheels are disposed on two sides of the vehicle body respectively and the wheels comprise rims matched with tracks. One end of the connection device is connected to a bottom of the vehicle body and the other end of the connection device is connected to the wheels. The connection device comprises an elastic member and the elastic member is configured to be deformed under the gravity of the vehicle body to adjust distances between the rims of the two groups of wheels. Under the gravity of the vehicle body, the track travelling device of the present disclosure can closely attach the rims of the wheels to side surfaces of the tracks, thereby preventing the vehicle body from swaying side to side. Meanwhile, the connection device comprises the elastic member for adaption to a change of a track gauge.