Abstract: An image pre-processing method and an image processing apparatus for a fundoscopic image are provided. A region of interest (ROI) is obtained from a fundoscopic image to generate a first image. The ROI is focused on an eyeball in the fundoscopic image. A smoothing process is performed on the first image to generate a second image. A value difference between neighboring pixels in the second image is increased to generate a third image.

Abstract: A backlight module including a light guide plate, a light source, a diffuse reflector and a light-splitting film is provided. The light guide plate has a light incident surface, and a light-emitting surface and a bottom surface which are respectively connected to the light incident surface and opposite to each other. The light source is disposed on one side of the light incident surface of the light guide plate. The diffuse reflector is disposed on one side of the bottom surface of the light guide plate. The light-splitting film is disposed between the light guide plate and the diffuse reflector. The light-splitting film has a substrate and a plurality of first optical microstructures disposed on one side of the substrate. An extending direction of the first optical microstructures intersects with the light incident surface of the light guide plate. A display apparatus using the backlight module is also provided.

Abstract: Disclosed are a power supply circuit, power supply method, audio power amplifier and integrated circuit. The power supply circuit applied to an audio power amplifier may include: a measurement sub-circuit used for measuring a voltage of an audio signal input to the audio power amplifier, which corresponds to output power of the audio power amplifier; a control sub-circuit used for determining a target reference voltage range according to a voltage range of the audio signal within a predetermined time period, which is selected from a plurality of predetermined reference voltage ranges, wherein the voltage range of audio signal within the predetermined time period is included in the target reference voltage range; and a power supply sub-circuit used for generating, based on the target reference voltage range, a power supply voltage that matches the output power of the audio power amplifier, so as to supply power to the audio power amplifier.

Abstract: An anti-peep light source module includes a light source module and a viewing angle switching module. The light source module has a light source and a light guide plate (LGP) and has light emitting elements arranged along a first direction. The viewing angle switching module is located on a transmission path of an illumination beam of the light source module and includes a viewing angle limiting element and a viewing angle adjusting element configured to change a viewing angle of the illumination beam. The viewing angle limiting element has a grating structure and is located between the viewing angle adjusting element and the light source module. an included angle between an extension direction of the grating structure and the first direction is within a range from 88 degrees to 92 degrees. The anti-peep light source module and A display device achieve favorable optical performance, user experience, and production yield.

Abstract: In an approach to augmenting a caption dataset by leveraging a denoising autoencoder to sample and generate additional captions from the ground truth captions, one or more computer processors generate a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions. The one or more computer processors calculate an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric. The one or more computer processors train a caption model with the generated plurality of new captions and associated calculated weights.

Abstract: In an approach to improving the image captioning performance of low-resource languages by leveraging multimodal inputs, one or more computer processors encode an image utilizing an image encoder, wherein the image is contained within a triplet comprising the image, one or more high-resource captions, and one or more low-resource captions. The one or more computer processors generate one or more high-resource captions utilizing the encoded image and the triplet inputted into a high-resource decoder. The one or more computer processors encode the one or more generated high-resource captions utilizing a high-resource encoder. The one or more computer processors add adaptive cycle consistency constraints on a set of calculated attention weights associated the triplet. The one or more computer processors generate one or more low-resource captions by simultaneously inputting the encoded image, the encoded high-resource caption, and the triplet into a trained low-resource decoder.

Abstract: Non-limiting examples of the present disclosure pertain to management over the type of notification that is generated and presented to a user relative to an occurrence of a message notification. Processing is executed that evaluates a relevance of a content update to content that is already presented to a user through host application/service endpoints. This helps determine how to present a graphical user interface (GUI) notification as well as link that GUI notification to existing communications that are identified as contextually relevant. Trained artificial intelligence modeling is applied that executes relevance analysis to aid generation of a GUI notification.

Abstract: A device and a method for acquiring brain electrical signal. The device for acquiring brain electrical signal includes a plurality of electrodes and an electrode base. The electrode base is provided with a plurality of screw holes, and each of the plurality of electrodes passes through a corresponding screw hole of the plurality of screw holes.

Abstract: Non-limiting examples of the present disclosure pertain to management over the type of notification that is generated and presented to a user relative to an occurrence of a message notification. Processing is executed that evaluates a relevance of a content update to content that is already presented to a user through host application/service endpoints. This helps determine how to present a graphical user interface (GUI) notification as well as link that GUI notification to existing communications that are identified as contextually relevant. Trained artificial intelligence modeling is applied that executes relevance analysis to aid generation of a GUI notification.

Abstract: In an approach to augmenting caption datasets, one or more computer processors sample a ratio lambda from a probability distribution based on a pair of datapoints contained in a dataset, wherein each datapoint in the pair of datapoints comprises an image and an associated caption; extend the dataset by generating one or more new datapoints based on the sampled ratio lambda for each pair of datapoints in the dataset, wherein the sampled ratio lambda incorporates an interpolation of features associated with the pair of datapoints into the generated one or more new datapoints; identify one or more objects contained within a subsequent image utilizing an image model trained utilizing the extended dataset; generate a subsequent caption for one or more identified objects contained within the subsequent image utilizing a language generating model trained utilizing the extended dataset.

Abstract: In an approach to augmenting a caption dataset by leveraging a denoising autoencoder to sample and generate additional captions from the ground truth captions, one or more computer processors generate a plurality of new captions utilizing an autoencoder fed with one or more noisy captions, wherein the autoencoder is trained with a dataset comprising a plurality of ground truth captions. The one or more computer processors calculate an importance weight for each new caption in the plurality of generated new captions as compared to a plurality of associated ground truth captions based on a consensus metric. The one or more computer processors train a caption model with the generated plurality of new captions and associated calculated weights.

Abstract: In an approach to improving the image captioning performance of low-resource languages by leveraging multimodal inputs, one or more computer processors encode an image utilizing an image encoder, wherein the image is contained within a triplet comprising the image, one or more high-resource captions, and one or more low-resource captions. The one or more computer processors generate one or more high-resource captions utilizing the encoded image and the triplet inputted into a high-resource decoder. The one or more computer processors encode the one or more generated high-resource captions utilizing a high-resource encoder. The one or more computer processors add adaptive cycle consistency constraints on a set of calculated attention weights associated the triplet. The one or more computer processors generate one or more low-resource captions by simultaneously inputting the encoded image, the encoded high-resource caption, and the triplet into a trained low-resource decoder.

Abstract: In an approach to augmenting caption datasets, one or more computer processors sample a ratio lambda from a probability distribution based on a pair of datapoints contained in a dataset, wherein each datapoint in the pair of datapoints comprises an image and an associated caption; extend the dataset by generating one or more new datapoints based on the sampled ratio lambda for each pair of datapoints in the dataset, wherein the sampled ratio lambda incorporates an interpolation of features associated with the pair of datapoints into the generated one or more new datapoints; identify one or more objects contained within a subsequent image utilizing an image model trained utilizing the extended dataset; generate a subsequent caption for one or more identified objects contained within the subsequent image utilizing a language generating model trained utilizing the extended dataset.

Abstract: A method for controlling functions of an electronic device by a server includes establishing a communication connection between the server and the electronic device when a distance between the electronic device and the server is less than or equal to a preset value. Positioning information of the electronic device is acquired at every preset time when the electronic device enters a preset control area. A control mode of the electronic device is determined according to positioning information of the electronic device. Once a control signal is generated according to the control mode of the electronic device, status of an image-capturing device, and/or status of a microphone of the electronic device are controlled according to the control signal.

Abstract: A high-pressure water jet sheet strip sand-blasting descaling cleaning device, a cleaning line and a system are disclosed. The cleaning device includes a cleaning box (7), a washing box (18) and descaling nozzles (6), the descaling nozzles (6) face to the same width direction of a sheet strip (5) and are arranged in a gradient along a conveying direction of the cleaning box (7) to jet covering the sheet strip (5). The system includes the cleaning device, an abrasive collecting box, a high-pressure water supply system, an abrasive supply-recovery-separation system and a sewage recycling treatment system. The nozzles are slantwise arranged in a gradient along the conveying direction, and the neighboring nozzles do not interfere with each other, and are used for cleaning successively, so that a cleaning process is stable and controllable.

Abstract: A two-dimensional code, a two-dimensional code generation method, identification method, generation apparatus, and identification apparatus, and a storage medium are provided. The two-dimensional code includes a square module array including an array of m*m modules, a location detection pattern and a data information pattern. The location detection pattern determines a location of the two-dimensional code, the data information pattern is configured to carry data information, each module of the array of m*m modules is at least one of a first-type module and a second-type module, a color attribute of the first-type module is different from a color attribute of the second-type module, and m is at least one of 17, 18, 19, and 20.

Abstract: A method for controlling functions of an electronic device by a server includes establishing a communication connection between the server and the electronic device when a distance between the electronic device and the server is less than or equal to a preset value. Positioning information of the electronic device is acquired at every preset time when the electronic device enters a preset control area. A control mode of the electronic device is determined according to positioning information of the electronic device. Once a control signal is generated according to the control mode of the electronic device, status of an image-capturing device, and/or status of a microphone of the electronic device are controlled according to the control signal.

Abstract: Embodiments of this application disclose an object on which a two-dimensional code is disposed, a two-dimensional code generation method, identification method, generation apparatus, and identification apparatus, and a storage medium. The two-dimensional code includes a square module array of m*m modules. The square module array includes a location detection pattern and a data information pattern. The location detection pattern is used for determining a location of the two-dimensional code. The data information pattern is used for carrying data. m=17 or 18 or 19 or 20. Even if printed in a miniature area whose side length is 0.5 cm to 0.7 cm, a miniature two-dimensional code provided in the embodiments of this application can still be normally identified, and is suitable for being used in a small area application scenario such as an inner side of a bottle cap or a corner of an object.

Abstract: A position tracking system and a method thereof are provided. In the system, a power transmitting unit (PTU) provides power to a power receiving unit (PRU). The PRU chooses one of the PTUs adaptively to receive power, and broadcasts a broadcast information via a corresponding Bluetooth link. Three PTUs receive the broadcast information to obtain three pieces of position information respectively corresponding to the PRU respectively, and a position of the PRU is determined based on the position information of the three PTUs. Accordingly, positioning is implemented in the wireless charging technique by modifying software design.