Disclosed is an optimization method for constructing a target detection network. The method includes: obtaining high-quality visual media data, low-quality visual media data corresponding to the high-quality visual media data, and a corresponding true label, extracting a first backbone network side output feature generated by a preset prior network for the high-quality visual media data, and extracting a second backbone network side output feature generated by a preset target detection network to be trained for the low-quality visual media data; constructing a feature correlation loss, a salient target position loss and a salient prediction loss based on the first backbone network side output feature, the second backbone network side output feature, and the true label; and optimizing the preset target detection network to be trained based on the feature correlation loss, the salient target position loss, and the salient prediction loss to obtain the target detection network.

An electronic apparatus is disclosed. The electronic apparatus includes: a memory storing an input image, and a processor configured to: apply a filter to the input image to identify the input image as a plurality of areas, apply a first low-frequency filter to a first area among the plurality of areas, and apply a second low-frequency filter to a second area among the plurality of areas to perform downscaling, wherein a cut-off frequency of the second low-frequency filter is configured to be higher than a cut-off frequency of the first low-frequency filter.

Apparatuses and methods associated with preserving authentication under item change are disclosed herein. In embodiments, acquiring digital image data of an image of at least a portion of a target physical object; extracting features from the image data to form a digital fingerprint; querying the database system to seek a matching record based on the digital fingerprint; based on an amount of difference between the digital fingerprint and a stored digital fingerprint of the database, update the database system to output a new indication of a new match to the physical object for any new samples that are not matchable to the stored digital fingerprint within a first predetermined similarity threshold provided the new samples are matchable to the digital fingerprint within a second predetermined similarity threshold. Other embodiments may be disclosed or claimed.

A data store to store and access digital records is provided, and a key object record is initialized in the data store to store data associated with a physical key object. A digital fingerprint of the physical key object is stored in the key object record. Another digital record is created in the data store that is not the key object record. The digital record is linked to the digital fingerprint of the physical key object. The linking is arranged to provide secure control access to the linked digital record. A tendered access key is received via a programmatic interface or user interface, and the data store is queried based on the tendered access key to identify a matching digital fingerprint of a key object. In a case that the querying identifies the matching digital fingerprint of the key object within a prescribed level of confidence, access to the linked digital record secured by the key object is granted.

Embodiments of this application disclose a terminal device capability transmission method, apparatus, and system. A terminal device reports, to a network device, capability information used to indicate a channel state information CSI reporting capability of the terminal device. The capability information is associated with a quantity, supported by the terminal device in a time-domain unit, of ports of pilots used for CSI measurement, and is used to enable the network device receiving the capability information to learn the CSI reporting capability of the terminal device, thereby determining a CSI measurement configuration of the terminal device.

An image processing method implemented by a computer includes extracting feature points from captured images that are sequentially generated by an image capture device and include at least a first captured image and a second captured image generated prior to the first captured image, determining whether the number of feature points extracted from the first captured image exceeds a threshold value, and specifying a location of the first captured image relative to the second captured image upon determining that the number of the feature points extracted from the first captured image is below the threshold value.

Systems and techniques for a design-aware image search are described. The design-aware image search techniques described herein capture a design on an item to determine additional items with similar or matching designs. An image of the item is used to create an edge image of a design, and shape descriptors are generated describing features of the edge image. These shape descriptors are compared to shape descriptors associated with other images to locate images that have similar or matching designs as compared with the input image. The design-aware image search system may uses these relationships to generate a search result with images or products having a design similar to the design on the input image.

Systems and methods are described herein for providing content item recommendations based on a video. Using feature vectors corresponding to at least one frame of a video (e.g., generated based on texture and shape intensity of a frame), a recommendation system improves content recommendation using analytic and quantitative characteristics derived from a frame of a content item rather than merely manually labeled bibliographic data (e.g., a genre or producer). The recommendation system may generate a feature vector based on a texture, a shape intensity (e.g., generated from a Generalized Hough Transform), and temporal data corresponding to at least one frame of a video. The feature vector is analyzed using a machine learning model (e.g., a neural network) to produce a machine learning model output. The recommendation system causes a recommended content item to be provided based on the machine learning model output.

Systems and methods for identifying an object and presenting additional information about the identified object are provided. The techniques of the present invention can allow the user to specify modes to help with identifying objects. Furthermore, the additional information can be provided with different levels of detail depending on user selection. Apparatus for presenting a user with a log of the identified objects is also provided. The user can customize the log by, for example, creating a multi-media album.

According to one implementation, a system for performing automated salience assessment of pixel anomalies includes a computing platform having a hardware processor and a system memory storing a software code. The hardware processor is configured to execute the software code to analyze an image for a presence of a pixel anomaly in the image, obtain a salience criteria for the image when the analysis of the image detects the presence of the pixel anomaly, and classify the pixel anomaly as one of a salient anomaly or an innocuous anomaly based on the salience criteria for the image. The hardware processor is further configured to execute the software code to disregard the pixel anomaly when the pixel anomaly is classified as an innocuous anomaly, and to flag the pixel anomaly when the pixel anomaly is classified as a salient anomaly.