Abstract: A method for optimizing a deep learning operator, includes: calling a method of reading an image object to read target data from an L1 cache of an image processor to the processor in response to detecting the target data in the L1 cache, performing a secondary quantization operation on the target data in the processor to obtain an operation result and writing the operation result into a main memory of the image processor. The target data is fixed-point data obtained after performing a quantization operation on data to be quantized in advance and the data to be quantized is one of the following: float-point data of an initial network layer of the neural network model and fixed-point data outputted from a network layer previous to the current network layer.

Abstract: A method for detecting an occluded image. The method includes: after an image is captured by a camera, obtaining the image as an image to be detected; inputting the image to be detected into a trained occluded-image detection model, the occluded-image detection model is trained based on original occluded images and non-occluded images by using a trained data feature augmentation network; determining whether the image to be detected is an occluded image based on the occluded-image detection model; and outputting an image detection result.

Abstract: A method and a device for generating an emoticon are provided. A first expression tag list corresponding to a face image in a portrait is acquired by inputting the face image into an expression recognition model. Additionally, at least one label text corresponding to the face image is determined based on the first expression tag list and a correspondence between a preset text and a second expression tag list. Furthermore, an expression image corresponding to the portrait is determined, where the face image is a part of the expression image. Moreover, an emoticon is generated by labelling the expression image with the at least one label text.

Abstract: A method for video clip extraction includes: obtaining a video, and sampling the video to obtain N video frames, wherein N is a positive integer; inputting the N video frames to a pre-trained frame feature extraction model to obtain a feature vector of each video frame in N video frames; determining scores of the N video frames based on a pre-trained scoring model; and extracting target video clips from the video based on the scores of the N video frames.

Abstract: A method for controlling a device includes: collecting audio data where the device is located; determining whether each target frame of the audio data is a first type signal; in response to the target frame of the audio data being the first type signal, determining an acoustic event type represented by the first type signal; and controlling the device to execute control instructions corresponding to the acoustic event type.

Abstract: A method for detecting an audio signal, the method comprises: obtaining a speech segment and a non-speech segment of an audio signal to be detected, extracting a first audio feature of the speech segment and a second audio feature of the non-speech segment, detecting the first audio feature using a predetermined speech segment detection model to obtain a first detection score, detecting the second audio feature using a predetermined non-speech segment detection model to obtain a second detection score, and determining whether the audio signal belongs to a target audio based on the first detection score and the second detection score.

Abstract: A method for extracting a video clip includes obtaining a video and splitting the video into multiple clips. The multiple clips are input into a pre-trained scoring model, to obtain a score of each of the multiple clips. The scoring model is obtained by training based on data pairs including first clips and second clips, the data pairs including first clips and second clips are obtained based on labeled clips labeled with target attributes, and the target attributes include attributes that characterize clips as target clips or non-target clips. A target clip is extracted from the multiple clips based on the score of each of the multiple clips. An apparatus for extracting a video clip and a computer readable storage medium are also disclosed.

Abstract: Multiple images are continuously captured based on a photographing instruction. Multiple processed images without ghosts are obtained based on the multiple images. A fused image is obtained by inputting the multiple processed images without ghosts into an image fusion model. Noise of the fused image is lower than noise of any one of the multiple images, and a resolution of the fused image is higher than a resolution of any one of the multiple images. A target image is obtained and output by performing image enhancement processing on the fused image.

Abstract: A method for semantic recognition includes: in response to performing semantic analysis on information acquired by a terminal, a sentence to be processed is acquired. Word recognition is performed on the sentence to be processed, to obtain a plurality of words and part-of-speech information thereof. A target set update operation is determined with a pre-trained word processing model, according to a word to be processed in the set of words to be processed and part-of-speech information of the word to be processed. If a dependency relationship corresponding to the target set update operation is a first dependency relationship, through each of the plurality of preset set update operations, a respective dependency relationship of the word to be processed and a respective confidence level corresponding to the dependency relationship is determined, and a respective update of the set of words to be processed is performed.

Abstract: A method and device for behavior recognition based on line-of-sight estimation are provided. The method is as follows. An image is collected. It is detected whether the image includes a face. When the image includes a face, it is determined whether the face in the image is a face with closed eyes. When the face in the image is not with closed eyes, a line-of-sight direction angle of a target to which the face belongs, and a distance between the face and an image collecting unit, are computed. It is determined whether the target has a set behavior based on the line-of-sight direction angle and the distance.

Abstract: Aspects of the present disclosure relate to a text labeling method and device, and more specifically to the field of deep learning and solving the problem of low efficiency and low accuracy of a feature extraction process. The method can include processing input information to obtain word embedding representation information of the input information, performing dynamic convolution feature extraction on the word embedding representation information to obtain a classification result of each character from the word embedding representation information, and inserting a label into the input information based on the classification result.

Abstract: A speech interaction method includes: acquiring speech information of a user; determining a task list corresponding to the speech information, the task list comprising at least two ordered tasks; and for each task in the at least two ordered tasks, responsive to that a next task of a present task is a question-answer task, querying and sending response information of the next task to a user terminal before execution time of the next task arrives, such that the user terminal outputs the response information when the execution time of the next task arrives.

Abstract: A method and device for behavior recognition are provided. According the present disclosure, an image is collected. It is detected whether the image includes a face. When the image includes a face, it is determined whether the face in the image is a face with closed eyes. When the face in the image is not with closed eyes, a distance between the face and an image collecting unit is computed. A maximum field of vision of a target to which the face belongs is determined according to the distance and a parameter of the image collecting unit. It is determined, according to the maximum field of vision, whether the target has a set behavior.

Abstract: A method and device for processing information, and a storage medium is provided. The method is applied to an authorization proxy server, and includes receiving a first account information of a first vehicle-mounted terminal, determining, based on an associating record of a user account service, that a first account corresponding to the first account information is an authorized account that has been associated with the user account service, and authorizing the first account with a control right for controlling a device to be controlled.

Abstract: A method, apparatus, and a non-transitory computer-readable storage medium for processing a video are provided. A terminal determines a subject region of a video frame in a video and a background region. A target object is located in the subject region. The background region is a region of the video frame other than the subject region. The terminal overlays the subject region in at least one of a first video frame having the target object on at least one of a second video frame having the target object and generates a special effect frame including at least two subject regions in each of which the target object is located.

Abstract: A method for controlling headphones. The headphones include a feedback microphone and a speaker. The method includes: obtaining an ear canal audio signal by collecting an audio signal in an ear canal by the feedback microphone; obtaining an audio signal feature parameter by conducting feature extraction on the ear canal audio signal; determining a tightness level of the headphones in a current wearing status; generating an audio feature to be identified according to the audio signal feature parameter and the tightness level; inputting the audio feature to be identified into a preset interactive operation identification model, and outputting an identification result, where the identification result includes an interactive operation identifier; determining a control instruction corresponding to the interactive operation identifier; and controlling a playback status of the speaker according to the control instruction.

Abstract: A speech control method, for a terminal device, includes: receiving an input speech control instruction, obtaining a recognition result of the speech control instruction; searching for an execution object matching the recognition result step by step within a preset search range; and responding to the speech control instruction based on a search result; in which the preset search range at least includes any one of: a current interface of the terminal device when receiving the speech control instruction, at least one application currently running on the terminal device when receiving the speech control instruction, and a system of the terminal device.

Abstract: A sound processing method includes: determining a vector of a first residual signal according to a first signal vector and a second signal vector, the first signal vector including a first voice signal and a first noise signal input into the first microphone, the second signal vector including a second voice signal and a second noise signal input into the second microphone, and the first residual signal including the second noise signal and a residual voice signal; determining a gain function of a current frame according to the vector of the first residual signal and the first signal vector; and determining a first voice signal of the current frame according to the first signal vector and the gain function of the current frame.

Abstract: A delay adjustment method is applied to a first electronic device, and the first electronic device is any device in a distributed device network. The method includes: a starting delay of a wakeup engine of at least one second electronic device in the network is acquired; it is determined whether a starting delay of a wakeup engine of the first electronic device needs to be adjusted according to the starting delay of the wakeup engine of the first electronic device and the starting delay of the wakeup engine of the at least one second electronic device; and responsive to that the starting delay of the first electronic device needs to be adjusted, the starting delay of the wakeup engine is adjusted, and wakeup information interaction is performed with the at least one second electronic device based on the adjusted starting delay of the wakeup engine.

Abstract: Aspects of the disclosure provide a responding method and device, an electronic device and a storage medium. The method is applied to a first electronic device including an audio acquisition component and an audio output component. The method can include acquiring a voice signal through the audio acquisition component, determining whether to respond to the voice signal, and responsive to determining to respond to the voice signal, outputting a first sound signal by the audio output component, the first sound signal being configured to notify at least one second electronic device that the first electronic device responds to the voice signal. In such a manner, an electronic device, responsive to determining to respond to a voice signal, outputs a sound signal to prevent other electronic device(s) from responding to the voice signal, so that competitions between electronic devices are reduced and a user experience is improved.