Embodiments of the present invention provide a model generation method, an image classification method, a controller, and an electronic device. The model generation method comprises: constructing a convolutional neural network model for image classification, and dividing the convolutional neural network model into N modules in sequence, wherein each module comprises multiple adjacent layers in the neural network model, and N is an integer greater than 1; based on unlabeled training data, training first to (N-1)-th module to obtain parameters and models of the first module to the (N-1)-th module; and cascading the trained first to (N-1)-th modules with N-th module, and training the cascaded N modules by using labeled training data, to obtain the parameters and models of the modules. A high-precision convolutional neural network model can be obtained without the need to label a large amount of training data, and the labor and time required for labeling the training data are saved.
G06V 10/764 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
G06V 10/774 - Generating sets of training patternsBootstrap methods, e.g. bagging or boosting
G06V 10/82 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
2.
MODEL GENERATION METHOD, OBJECT DETECTION METHOD, CONTROLLER AND ELECTRONIC DEVICE
This invention provides a model generation method, an object detection method, a controller, and an electronic device. The model generation method comprises: constructing a convolutional neural network model used for multi-scale object detection, and dividing the convolutional neural network model into a plurality of modules, the plurality of modules comprising a feature extraction module and a plurality of detection head modules of different scales; using unlabeled training data to pre-train the feature extraction module to obtain parameters and models of the feature extraction module; and connecting the trained feature extraction module to the plurality of detection head modules respectively, and using labeled training data to train a plurality of the modules which have been connected, to obtain parameters and models of the modules. A high-precision convolutional neural network model can be obtained without the need to label a large amount of training data, and the labor and time required for labeling the training data are saved.
G06V 10/82 - Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
G06V 10/77 - Processing image or video features in feature spacesArrangements for image or video recognition or understanding using pattern recognition or machine learning using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]Blind source separation
G06V 10/774 - Generating sets of training patternsBootstrap methods, e.g. bagging or boosting
3.
MODEL GENERATION METHOD, IMAGE CLASSIFICATION METHOD, CONTROLLER, AND ELECTRONIC DEVICE
Embodiments of the present invention provide a model generation method, an image classification method, a controller, and an electronic device. The model generation method comprises: constructing a convolutional neural network model for image classification, and sequentially dividing the convolutional neural network model into N modules, wherein each module comprises multiple adjacent layers in the neural network model, and N is an integer greater than 1; on the basis of unlabeled training data, training a first module to an (N-1)th module so as to obtain parameters and models of the first module to the (N-1)th module; and cascading the trained first to (N-1)th modules with an Nth module, and training the cascaded N modules by using labeled training data, so as to obtain the parameters and models of the modules. A high-precision convolutional neural network model can be obtained without the need to label a large amount of training data, and the labor and time required for training data labeling are saved.
A model generation method, an object detection method, a controller, and an electronic device. The model generation method comprises: constructing a convolutional neural network model used for multi-scale object detection, and dividing the convolutional neural network model into a plurality of modules, the plurality of modules comprising a feature extraction module and a plurality of detection head modules of different scales (101); pre-training the feature extraction module by using unlabeled training data to obtain a parameter and model of the feature extraction module (102); and connecting the trained feature extraction module to the plurality of detection head modules separately, and by using labeled training data, training the plurality of modules which have been connected, so as to obtain a parameter and a model of each module (103). A high-precision convolutional neural network model can be obtained without the need of a large amount of labeled training data, saving manpower and time required for labeling trained data.
Disclosed is a curtain. In the present invention, the curtain comprises: a curtain body, the curtain body having a top part and a bottom part arranged opposite to the top part; and a control apparatus which is arranged at the bottom part of the curtain body. The control apparatus has a housing, a driving module, and an actuator module and power supply module that are connected to the driving module; the driving module, the actuator module and the power supply module are arranged in the housing, and the power supply module is electrically connected to the driving module; and the driving module is used to drive the actuator module to open and close the curtain body in the direction from the top part thereof to the bottom part. Compared to the prior art, the structure is simple and ingenious, the installation and maintenance are convenient and simple, and the scope of application is wide.
The invention discloses a smart loudspeaker, wherein the smart loudspeaker includes a voice input module, a language recognition module and at least two voice assistants, and the language recognition module receives a voice information from the voice input module and determines the language category based on the voice information and activates the voice assistant corresponding to the language category.
G10L 17/02 - Preprocessing operations, e.g. segment selectionPattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal componentsFeature selection or extraction
H04L 12/28 - Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
7.
Intelligent switch panel and intelligent remote control switch system
The invention discloses a smart switch panel and a smart remote control switch system. The smart switch panel includes a housing, an actuating device arranged in the housing and a microprocessor arranged in the housing to control the actuating device, and the actuating device includes a rotating rod and a driving member, the rotating rod cooperates with the driving member and pushes the driving member to move in a direction parallel to the rotating rod when the rotating rod rotates, so as to drive the switch to switch when the smart switch panel is mounted on the switch.
H02H 3/26 - Emergency protective circuit arrangements for automatic disconnection directly responsive to an undesired change from normal electric working condition, with or without subsequent reconnection responsive to difference between voltages or between currentsEmergency protective circuit arrangements for automatic disconnection directly responsive to an undesired change from normal electric working condition, with or without subsequent reconnection responsive to phase angle between voltages or between currents
G08C 17/02 - Arrangements for transmitting signals characterised by the use of a wireless electrical link using a radio link
H01H 3/26 - Power arrangements internal to the switch for operating the driving mechanism using dynamo-electric motor
A text input system based on a speech recognition device, the text input system comprising: a speech recognition device, a text input core module and an application terminal, wherein the speech recognition device is in communication connection with the text input core module, and the text input core module is in communication connection with the application terminal. The text input core module comprises an input module and an output module, wherein the input module is in communication connection with the output module. The input module comprises an input control unit and an input unit, wherein the input control unit is in communication connection with the input unit. The output module comprises an output control unit and an output unit, wherein the output control unit is in communication connection with the output unit. The input module is in communication connection with the output module by means of the input control unit and the output control unit. By means of the text input system based on a speech recognition device, and a method therefor, a speech-to-text input means can be used on an application terminal, without the installation of any software.
Provided is a curtain (100), comprising: a curtain body (2), the curtain body (2) having a top (21) and a bottom (22) arranged opposite to the top (21); a tensile member (3), at least part of the curtain body (2) running through the top (21) of the curtain body (2) in the direction of the top (21); a control apparatus (20), arranged at the bottom (22) of the curtain body (2), to drive the tensile member (3) to open and close the curtain body (2) in the direction from its top (21) to its bottom (22). In comparison with the prior art, the invention has a simple and ingenious structure, is easy to install and maintain, and has a broad range of applications.
An intelligent switch panel (100) used for cooperating with a switch (200). The switch (200) is provided with multiple buttons (201, 202, 203). The intelligent switch panel (100) comprises a housing (10), an actuating device (30), a driving device (20), and a microprocessor. The actuating device (30) is disposed in the housing (10) to actuate the switch (200). The driving device (20) is disposed in the housing (10) to drive the actuating device (30) for translation to cooperate with the different buttons (201, 202, 203) of the switch (200) and to drive the actuating device (30) to rotate to toggle the buttons (201, 202, 203) of the switch (200). The microprocessor is disposed in the housing (10) to control the driving device (20).
Disclosed is a smart speaker, characterized in that the smart speaker comprises a voice input module, a language recognition module and at least two voice assistants, wherein the language recognition module receives voice information from the voice input module, determines a language category according to the voice information, and activates a voice assistant corresponding to the language category.
Disclosed are an intelligent switch panel and an intelligent remote control switch system. The intelligent switch panel comprises a shell, an actuating device provided in the shell, and a microprocessor. The microprocessor is provided in the shell to control the actuating device. The actuating device comprises a rotating rod and a driving member. The rotating rod matches the driving member, and when the rotating rod rotates, the driving member is driven to move in the direction parallel to the rotating rod, to drive a switch for switching when the intelligent switch panel is mounted on the switch.
Disclosed are a smart curtain control device and a smart curtain. The smart curtain control device comprises a base (1). The base (1) is used for being mounted on window glass or a wall. The smart curtain control device further comprises a control assembly (2), the control assembly (2) being slidably mounted on the base (1). The control assembly (2) comprises a housing (21), a controller and a driving mechanism (22). The controller is mounted in the housing (21) and is electrically connected to the driving mechanism (22) to control the operation of the driving mechanism (22). The driving mechanism (22) is mounted on the housing (21) and is provided with a rotating disc (222). The rotating disc (222) is used for being connected to and driving a pull rope of a curtain. The smart curtain control device can be used in combination with an existing mounted curtain to enable automatic control of the curtain, and is simple to mount and easy to use.
E06B 9/68 - Operating devices or mechanisms, e.g. with electric drive
G05B 19/418 - Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
A smart kitchen system (100) comprising a smart scale (10), a mobile application (30), a smart loudspeaker (20) and a cloud server (40), wherein the mobile application is in a signal communication with the smart scale and the cloud server by means of a wireless connection; and the smart loudspeaker interacts with a user by means of voice and is in a signal communication with the cloud server by means of a wireless connection so as to be associated with the smart scale. The smart kitchen system can access a wide range of recipes by means of voice commands, thereby helping a user to make food or beverages with little cooking knowledge.
G01G 19/40 - Weighing apparatus or methods adapted for special purposes not provided for in groups with provisions for indicating, recording, or computing price or other quantities dependent on the weight
G09B 5/06 - Electrically-operated educational appliances with both visual and audible presentation of the material to be studied
15.
REMOTE CONTROL METHOD AND SERVER THEREFOR, DEVICE AND STORAGE MEDIUM
The present application relates to the field of communications. Disclosed are a remote control method and a server therefor, a device and a storage medium. The remote control method in the present application comprises: a server receiving control instruction information; the server determining, according to the received control instruction information, a physical execution mechanism group to be controlled; and the server sending instruction information to at least one physical execution mechanism in the determined physical execution mechanism group, such that the physical execution mechanism executes a corresponding operation, wherein the physical execution mechanism group comprises at least two physical execution mechanisms. According to the present application, a server can perform remote overall control of a group of physical execution mechanisms.
G05B 19/418 - Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
patents
