Embodiments may provide techniques that provide identification of images that can provide reduced resource utilization due to reduced sampling of video frames for visual recognition. For example, in an embodiment, a method of visual recognition processing may be implemented in a computer system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, the method comprising: coarsely segmenting video frames of video stream into a plurality of clusters based on scenes of the video stream, sampling a plurality of video frames from each cluster; determining a quality of each cluster, re-clustering the video frames of video stream to improve the quality of at least some of the clusters.

Aspects of the current disclosure include systems and methods for identifying an entity in a query image by comparing the query image with digital images in a database. In one or more embodiments, a query feature may be extracted from the query image and a set of candidate features may be extracted from a set of images in the database. In one or more embodiments, the distances between the query feature and the candidate features are calculated. A feature, which includes a set of shortest distances among the calculated distances and a distribution of the set of shortest distances, may be generated. In one or more embodiments, the feature is input to a trained model to determine whether the entity in the query image is the same entity associated with one of the set of shortest distances.

A method performed by an apparatus is described. The method includes receiving map data that is based on first image data, second image data, and a similarity metric. The first image data can be received from a first vehicle and represent an object. The second image data can be received from a second vehicle and represent the object. The similarity metric can be associated with the object represented in the first image data and the object represented in the second image data. The method can also include storing, by a vehicle, the received map data and localizing the vehicle based on the stored map data.

The present disclosure provides various approaches for smart area monitoring suitable for parking garages or other areas. These approaches may include ROI-based occupancy detection to determine whether particular parking spots are occupied by leveraging image data from image sensors, such as cameras. These approaches may also include multi-sensor object tracking using multiple sensors that are distributed across an area that leverage both image data and spatial information regarding the area, to provide precise object tracking across the sensors. Further approaches relate to various architectures and configurations for smart area monitoring systems, as well as visualization and processing techniques. For example, as opposed to presenting video of an area captured by cameras, 3D renderings may be generated and played from metadata extracted from sensors around the area.

A method for clustering based on unsupervised learning according to an embodiment of the invention enables clustering for newly generated patterns and is robust against noise, and does not require tagging for training data. According to one or more embodiments, noise is accurately removed using three-dimensional stacked spatial auto-correlation, and multivariate spatial probability distribution values and polar coordinate system spatial probability distribution values are used as learning features for clustering model generation, making them robust to noise, rotation, and fine unusual shapes. In addition, clusters resulting from clustering are classified into multi-level clusters, and stochastic automatic evaluation of normal/defect clusters is possible only with measurement data without a label.

An electronic device and a method for controlling thereof are provided. A method for controlling an electronic device according to the disclosure includes obtaining a plurality of images for performing clustering, obtaining a plurality of target areas corresponding to each of the plurality of images, obtaining a plurality of feature vectors corresponding to the plurality of target areas, obtaining a plurality of central nodes corresponding to the plurality of feature vectors, obtaining neighbor nodes associated with each of the plurality of central nodes, obtaining a subgraph based on the plurality of central nodes and the neighbor nodes, identifying the connection probabilities between the plurality of central nodes of the subgraph and the neighbor nodes of each of the plurality of central nodes based on a graph convolutional network, and clustering the plurality of target areas based on the identified connection probabilities.

The present invention provides a processing apparatus (10) including an inter-face-image-group determination unit (11) that determines whether a similarity score between a first representative face image in a first face image group and a second representative face image in a second face image group satisfies a first condition, an intra-face-image-group determination unit (12) that determines, for each of the first face image group and the second face image group, whether a second condition defining a relationship between the representative face image and another image in the face image group is satisfied, based on a similarity score between the first representative face image and each of other face images in the first face image group and a similarity score between the second representative face image and each of other face images in the second face image group, and a processing unit (13) that associates a same person identifier (ID) with a plurality of face images included in the first face image group and the second face image group when it is determined that the first condition is satisfied and that the first face image group and the second face image group each satisfy the second condition.

Techniques are disclosed for incremental agglomerative clustering of data, including but not limited to digital image data. Fewer than all of a plurality of existing digital image fingerprints are sampled from a first hierarchical data cluster of digital image fingerprints stored in a data storage device, the first hierarchical data cluster excluding a new digital image fingerprint. The new digital image fingerprint and the existing digital image fingerprints sampled from the first hierarchical data cluster are clustered to produce a second hierarchical data cluster of digital image fingerprints, the second hierarchical data cluster including the new digital image fingerprint. If a majority of the existing digital image fingerprints in the first hierarchical data cluster match the new digital image fingerprint, then the second hierarchical data cluster is mapped to the first hierarchical data cluster based on the determination.

A computer model is trained to account for data samples in a high-dimensional space as lying on different manifolds, rather than a single manifold to represent the data set, accounting for the data set as a whole as a union of manifolds. Different data samples that may be expected to belong to the same underlying manifold are determined by grouping the data. For generative models, a generative model may be trained that includes a sub-model for each group trained on that group's data samples, such that each sub-model can account for the manifold of that group. The overall generative model includes information describing the frequency to sample from each sub-model to correctly represent the data set as a whole in sampling. Multi-class classification models may also use the grouping to improve classification accuracy by weighing group data samples according to the estimated latent dimensionality of the group.

A method may include clustering form elements into line objects and columns of a table of a structured representation by applying a trained multi-dimensional clustering model to spatial coordinates of the form elements, and assigning a table header line type to a table header line object of the line objects based on a spatial coordinate of the table header line object relative to a spatial coordinate of a topmost table data line object of the line objects, and a determination that a number of columns of the table header line object is within a threshold of a number of columns of the topmost table data line object. The topmost table data line object may be assigned a table data line type. The method may further include presenting the structured representation to a user.