Abstract: A design management apparatus is provided for presenting a template used to create a design. The design management apparatus may include a calculator that receives multiple templates having multiple character strings as predetermined samples and calculates a score for each character string in each template, where the score concerns a subject of the template and corresponds to a size of the character string. Moreover, the design management apparatus may receive information indicating a search subject to output the template including the character string which corresponds to the search subject and the score of which meets an output condition.

Abstract: Systems and methods for estimating the pose of a textureless object are disclosed. A method to estimate a pose of a textureless object includes obtaining, by a processing device, a single image of the textureless object. The pose of the textureless object can be inferred from the single image. The method further includes generating, by the processing device, a three dimensional model of the textureless object from a plurality of viewpoints and a plurality of scales obtained from image data of the textureless object, matching, by the processing device, the single image with a discretized render of the three dimensional model via a multi-level illumination invariant tree structure to obtain an alignment of the single image with the discretized render, and estimating, by the processing device, the pose of the textureless object based on the alignment.

Abstract: Systems and methods for addressing missing features in models are provided. In some implementations, a model configured to indicate likelihoods of different outcomes is accessed. The model includes a respective score for each of a plurality of features, and each feature corresponds to an outcome in an associated context. It is determined that the model does not include a score for a feature corresponding to a potential outcome in a particular context. A score is determined for the potential outcome in the particular context based on the scores for one or more features in the model that correspond to different outcomes in the particular context. The model and the score are used to determine a likelihood of occurrence of the potential outcome.

Abstract: A method for a processing device to determine whether to assign a data item to at least one cluster of data items is disclosed. The processing device may identify a signature of the data item, the signature including a set of elements. The processing device derive a first size value of the number of elements of the identified signature based on a set of size values of signatures that includes a maximum size value representing the largest number of elements in a signature. The processing device may derive a second size value of the number of elements of a second signature that is similar to the identified signature based on the set of size values of signatures.

Abstract: The trend of analyzing big data in artificial intelligence requires more scalable machine learning algorithms, among which clustering is a fundamental and arguably the most widely applied method. To extend the applications of regular vector-based clustering algorithms, the Discrete Distribution (D2) clustering algorithm has been developed for clustering bags of weighted vectors which are well adopted in many emerging machine learning applications. The high computational complexity of D2-clustering limits its impact in solving massive learning problems. Here we present a parallel D2-clustering algorithm with substantially improved scalability. We develop a hierarchical structure for parallel computing in order to achieve a balance between the individual-node computation and the integration process of the algorithm. The parallel algorithm achieves significant speed-up with minor accuracy loss.

Abstract: In an example embodiment, a method, apparatus and computer program product are provided. The method includes facilitating receipt of an image of a scene and determining a graph based on connecting nodes of the image. The nodes are either pixels or superpixels of the image. The graph is determined by determining one or more connections of a node to one or more nodes belonging to a pre-defined image region around the node in the image. The connections are associated with edge weights that are determined based on at least one of similarity parameters and spatial distances between the node and the one or more nodes. The method includes determining disparity values at the nodes of the image based at least on performing tree based aggregation of a cost volume on the graph, where the cost volume is associated with the image and at least one view image of the scene.

Abstract: According to one embodiment, an electronic apparatus includes a display processor. The display processor is configured to display one or more first strokes on a screen. The display processor is configured to display a plurality of candidates each indicative of one or more handwritten characters in a handwritten document, the plurality of candidates retrieved from the handwritten document by using the one or more first strokes, and is configured to arrange the plurality of candidates in an order determined by using number of strokes of each of the plurality of candidates and to display the plurality of arranged candidates.

Abstract: An automaton hardware engine employs a transition table organized into 2n rows, where each row comprises a plurality of n-bit storage locations, and where each storage location can store at most one n-bit entry value. Each row corresponds to an automaton state. In one example, at least two NFAs are encoded into the table. The first NFA is indexed into the rows of the transition table in a first way, and the second NFA is indexed in to the rows of the transition table in a second way. Due to this indexing, all rows are usable to store entry values that point to other rows.

Abstract: This disclosure describes techniques enabling character-string completion. These techniques present characters and character strings responsive to a selected prior character and enable selection of one of these characters or character strings quickly, easily, and/or accurately.

Abstract: A method for identifying a best item among a plurality of respective different items includes selecting a current best score, identifying a set of attributes associated with each of the items, and for each of the sets of attributes: determining a best possible score for a subset of the set of attributes, if the best possible score is greater than the current best score, determining an actual score for the set of attributes, and if the actual score for the set of attributes is greater than the current best score, setting the current best score as the actual score for the set of attributes. After both of the determining steps and the setting step for each of the sets of attributes, the item associated with the current best score is identified.

Abstract: An NFA (Non-deterministic Finite Automaton) circuit includes a hardware byte characterizer, a first matching circuit (performs a TCAM match function), a second matching circuit (performs a wide match function), a multiplexer that outputs a selected output from either the first or second matching circuits, and a storage device. N data values stored in first storage locations of the storage device are supplied to the first matching circuit as an N-bit mask value and are simultaneously supplied to the second matching circuit as N bits of an N+O-bit mask value. O data values stored in second storage locations of the storage device are supplied to the first matching circuit as the O-bit match value and are simultaneously supplied to the second matching circuit as O bits of the N+O-bit mask value. P data values stored in third storage locations are supplied onto the select inputs of the multiplexer.

Abstract: Provided is a system and method for object detection and tracking in a video stream. Frames of the video stream are divided into regions of interest and a probability that the region contains at least a portion of an object to be tracked is calculated for each region of interest. The regions of interest in each frame are then classified based on the calculated probabilities. A region of interest (RI) frame is then constructed for each video frame that reports the classification of regions of interest in the video frame. Two or more RI frames are then compared in order to determine a motion of the object. Also provided is a system executing the presently described method, as well as a device including the system. The device may be for example, a portable computer, a mobile telephone, or an entertainment device.

Abstract: Systems, methods, and non-transitory computer-readable media can acquire historical data including a plurality of features associated with known legitimate activities and with known illegitimate activities. A machine learning technique can be applied to the historical data to gain information about the plurality of features associated with the known legitimate activities and with the known illegitimate activities. A decision tree can be generated based on at least a portion of the information about the plurality of features. A node in the decision tree that satisfies specified precision criteria can be identified. A rule can be created based on the node. One or more illegitimate activities can be identified based on the rule.

Abstract: A representative feature extraction system which selects a representative feature from an input data group includes: occurrence distribution memory means for memorizing an occurrence distribution with respect to feature quantities assumed to be input; evaluation value calculation means for calculating, with respect to each of data items in the data group, the sum of distances to the other data items included in the data group based on the occurrence distribution, to determine an evaluation value for the data item; and data selecting means for selecting the data item having the smallest evaluation value as a representative feature of the data group.

Abstract: A computing device for use in decision tree computation is provided. The computing device may include a software program executed by a processor using portions of memory of the computing device, the software program being configured to receive user input from a user input device associated with the computing device, and in response, to perform a decision tree task. The computing device may further include a decision tree computation device implemented in hardware as a logic circuit distinct from the processor, and which is linked to the processor by a communications interface. The decision tree computation device may be configured to receive an instruction to perform a decision tree computation associated with the decision tree task from the software program, process the instruction, and return a result to the software program via the communication interface.

Abstract: A method executes software including a view hierarchy with a plurality of views which displays one or more views of the view hierarchy. The method executes software elements associated with a particular view, wherein each particular view includes event recognizers. Each event recognizer has an event definition based on sub-events, and an event handler that specifies an action for a target, and is configured to send the action to the target in response to an event recognition. The method detects a sequence of sub-events, and identifies one of the views of the view hierarchy as a hit view that establishes which views in the hierarchy are actively involved views. The method delivers a respective sub-event to event recognizers for each actively involved view, wherein each event recognizer for actively involved views in the view hierarchy processes the respective sub-event prior to processing a next sub-event in the sequence of sub-events.

Abstract: Systems and Methods for multi-modal or multimedia image retrieval are provided. Automatic image annotation is achieved based on a probabilistic semantic model in which visual features and textual words are connected via a hidden layer comprising the semantic concepts to be discovered, to explicitly exploit the synergy between the two modalities. The association of visual features and textual words is determined in a Bayesian framework to provide confidence of the association. A hidden concept layer which connects the visual feature(s) and the words is discovered by fitting a generative model to the training image and annotation words. An Expectation-Maximization (EM) based iterative learning procedure determines the conditional probabilities of the visual features and the textual words given a hidden concept class. Based on the discovered hidden concept layer and the corresponding conditional probabilities, the image annotation and the text-to-image retrieval are performed using the Bayesian framework.

Abstract: An information processing apparatus and method enables a pattern discriminator to learn. The apparatus establishes a branch structure appropriate for learning a discriminator having the branch structure without increasing processing time. The apparatus includes a preliminary learning unit to learn a preliminary discriminator for a respective one of a plurality of combinations of variations in variation categories in a discrimination target pattern. A branch structure determination unit is provided to perform discrimination processing using the preliminary discriminator and to determine a branch structure of a main discriminator based on a result of the discrimination processing. A main learning unit is included to learn the main discriminator based on the branch structure.

Abstract: A computer-implemented method performs foreground segmentation of an input image. The method receives a first foreground segmentation at a first resolution of the input image and determines a plurality of labelled seed points based on the first foreground segmentation of the input image. The method associates each of the plurality of pixels in the input image with one of the determined labelled seed points to obtain a second foreground/background segmentation of the input image, and performs foreground separation on the input image at a second resolution by classifying each of the segments of the second segmentation as one of foreground and background based on the label of the associated seed point.

Abstract: Disclosed is an image measuring apparatus including an image capture unit which captures an image of a measurement subject, an import unit which imports the image of the measurement subject which is captured by the image capture unit, a binarization unit which binarizes the image which is imported by the import unit, a contour detection unit which recognizes graphic information in the image which is binarized by the binarization unit and detects a contour of the graphic information, a corner detection unit which detects corners of the graphic information based on the contour which is detected by the contour detection unit, a setting unit which respectively sets edge detection tools on lines of the contour including the corners detected by the corner detection unit and a measurement unit which measures the graphic information by the edge detection tools which are set by the setting unit.

Abstract: Portable devices are equipped with a variety of technologies by which existing functionality can be improved, and new functionality can be provided. One claim recites a method comprising: receiving data representing imagery that depicts an object, the imagery captured by a portable device, the portable device comprising a camera and a microphone; determining one or more descriptors relating to the object in the imagery, the determining including collecting descriptors associated with other imagery or audio; processing the descriptors in discerning whether the imagery depicts an object that is likely of a first class or a second class or a third class, the processing being performed by one or more electronic processors configured to perform such act; and taking an action dependent on whether the imagery depicts an object that is likely of a first class or a second class or a third class. Of course, other claims and combinations are provided as well.

Abstract: Disclosed herein are sensors that include: (a) a circuit board that includes an electronic processor; (b) a plurality of radiation sources, each source being attached to the circuit board; and (c) a spectral detector attached to the circuit board, the spectral detector being configured to analyze radiation derived from one or more of the plurality of radiation sources. During use, the sensors are configured to be worn on a portion of a body of a subject. The electronic processor is configured to cause two or more of the plurality of radiation sources to direct incident radiation to the subject, to cause the spectral detector to analyze radiation from the subject, and to determine one or more properties of the subject based on the radiation from the subject. Methods of making and using these sensors are also disclosed.

Abstract: An information processing apparatus including an acquisition unit that acquires a plurality of image sets obtained by different acquisition methods, a generation unit that generates partial image sets by extracting corresponding partial images from respective images of the image sets, an assignment unit that assigns the partial image sets to a root node of a tree structure, a setting unit that sets, at each node of the tree structure, positions of a plurality of pixels for each partial image set assigned to the node, a determination unit that determines whether any one of the plurality of pixels in a. partial image obtained by a predetermined one of the different acquisition methods in each partial image set has an invalid value, and a selection unit that selects art acquisition method based on the determination.

Abstract: An image verification device that checks an input image obtained by photographing an object to be checked against a registered image database, wherein, in the registered image database, an amount of feature of an image obtained by photographing an object is registered as a registered image, and the registered image includes registered images registered with respect to a plurality of objects, has a verification score calculating unit that calculates a verification score serving as a score representing a degree of approximation between the objects indicated by the registered images and the object of the input image by using the amount of feature of the input image and the amounts of feature of the registered images, and a relative evaluation score calculating unit.

Abstract: In one embodiment, a method includes at time t2, determining a delta set of first objects representing a difference between a first set of first objects at time t1 and a second set of first objects at time t2; comparing the delta set of first objects with a set of second objects represented as a balanced k-dimensional tree; and identifying each second object whose content substantially matches content of at least one first object from the delta set of first objects.

Abstract: Embodiments of methods and apparatuses for searching contents, including structured search for atomic search expressions, including proximately associated atomic search expressions, are described herein. Embodiments may use tree structures (or more generally, graph structures), layout structures, and/or other information to capture within search results relevant content, include sub-document constituents, to reduce the incidence of false positives within search results, and/or to improve the accuracy of rankings within search results. Embodiments may use distance and/or scoring functions to generate scores for the structures to indicate relevance, including usage of local geometry, and linear iteration over portions of the content at a level to capture potential of a portion to influence other portions of the level, and influence received by a portion from the other portions of the level. Other embodiments may be described and claimed.

Abstract: Provided are an age estimation apparatus, an age estimation method, and an age estimation program capable of obtaining a recognition result closely matching the result perceived by human. An age estimation apparatus 10 for estimating an age of a person on image data includes a dimension compressor 11 for applying dimension compression to the image data to output low dimensional data; and an identification device 12 for estimating an age of a person on the basis of a learning result using a feature amount contained in the low dimensional data, wherein a parameter used for the dimension compression by the dimension compressor 11 and the feature amount used for age estimation by the identification device 12 are set on the basis of a result of an evaluation of a generalization capability using a weighting function that shows a degree of seriousness of an age estimation error for every age, and learning of the identification device 12 is performed on the basis of the weighting function.

Abstract: A camera system comprises an image capturing device and an object classification module connected to the image capturing device. The object classification module is operable to determine whether or not an object in an image is a member of an object class. The object classification module includes multiple decision steps configured in a cascade configuration, wherein at least one of the multiple decision steps is operable to (a) accept an object as a member of the object class, (b) reject an object as a member of the object class, and (c) call on a next step to determine whether or not an object is a member of the object class.

Abstract: A method for automatically detecting and tracking multiple targets in a multi-camera surveillance zone and system thereof. In each camera view of the system only a simple object detection algorithm is needed. The detection results from multiple cameras are fused into a posterior distribution, named TDP, based on the Bayesian rule. This TDP distribution represents a likelihood of presence of some moving targets on the ground plane. To properly handle the tracking of multiple moving targets with time, a sample-based framework which combines Markov Chain Monte Carlo (MCMC), Sequential Monte Carlo (SMC), and Mean-Shift Clustering, is provided. The detection and tracking accuracy is evaluated by both synthesized videos and real videos. The experimental results show that this method and system can accurately track a varying number of targets.

Abstract: A computer-implemented method for automatically retrieving information regarding optical properties of a scattering medium including receiving a first digital image and a first image quality value associated with the first digital image and sharpness of an edge of the first digital image, producing an optimized image, transforming the optimized image into an optimized optical transfer function, receiving a second digital image and a second image quality value associated with the second digital image and sharpness of an edge of the second digital image, identifying either the first or second digital image as an optimized digital image, and transforming the optimized optical transfer function into an optimized value of the optical parameter.

Abstract: A method, system, and computer program product for detecting objects. Members of a general class of objects are searched for in a number of images. Members of a specific class of objects are searched for in a number of regions in the number of images where the number of regions contains at least a portion of the members of the general class. A member in the members of the specific class is a potential threat to a rotorcraft.

Abstract: A method for recognition of a predetermined pattern in an image data set recorded by a device for recording of at least two electromagnetic frequency spectra is provided. A first difference value is formed for the image points of the selected area as a function of a difference between a data vector of a corresponding image point and a first reference data vector. A second difference value is formed for an image point of a selected area as a function of a difference between the data vector of this image point and a second reference data vector. A predetermined pattern is recognized when it is determined at least one pattern correlation quantity is below a predetermined threshold value and a local minimum is present.

Abstract: An activity recognition method, for recognizing continuous activities of several moving objects in the foreground of a video, includes: capturing and processing a training video to get a contour of a moving object; extracting a minimum bounding box of the contour in order to get parameters then transfer to feature vectors; constructing a decision tree model based on support vector machines (SVMs), for classifying the activities of the moving object according to the parameter and the feature vector of the training video; capturing and processing a testing video to get other parameters and using several formulas to generate feature vectors, and executing an algorithm for recognizing the activities of several moving objects in the foreground of the testing video. Said feature vectors are transformed from the parameters that in the testing and training videos. Via above descriptions, the method can recognize activities of foreground objects in the testing video.

Abstract: Methods and devices for optimizing classification tree evaluation for online and real-time characterization of tissue by using GPUs are disclosed that give a significant improvement in run time compared to data decomposition. The methods and devices use a parallel algorithm designed for single instruction, multiple data (SIMD) architectures commonly found in GPUs. Various optimizations and their effects are also disclosed.

Abstract: A set of training vectors may be identified. Each training vector may be mapped to either a male gender or a female gender, and each training vector may represent facial landmarks derived from a respective facial image. An input vector of facial landmarks may also be identified. The facial landmarks of the input vector may be derived from a particular facial image. A feature vector may containing a subset of the facial landmarks may be selected from the input vector. A weighted comparison may be performed between the feature vector and each of the training vectors. Based on a result of the weighted comparison, the particular facial image may be classified as either the male gender or the female gender.

Abstract: Methods, systems, and articles of manufacture for annotating of an image are disclosed. These include scoring the image using a plurality of trained classifiers, wherein each of the trained classifiers corresponds to at least one of a plurality of image groups clustered based upon image similarity, and wherein each image group is associated with a set of weighted labels; selecting one or more of the image groups based upon the scoring; aggregating one or more sets of weighted labels associated with the selected one or more image groups; and annotating the image using the aggregated one or more sets of weighted labels.

Abstract: A new tractable model solves labeling problems using regression tree fields, which represent non-parametric Gaussian conditional random fields. Regression tree fields are parameterized by non-parametric regression trees, allowing universal specification of interactions between image observations and variables. The new model uses regression trees corresponding to various factors to map dataset content (e.g., image content) to a set of parameters used to define the potential functions in the model. Some factors define relationships among multiple variable nodes. Further, the training of regression trees is scalable, both in the training set size and in the fact that the training can be parallelized. In one implementation, maximum pseudolikelihood learning provides for joint training of various aspects of the model, including feature test selection and ordering (i.e., the structure of the regression trees), parameters of each factor in the graph, and the scope of the interacting variable nodes used in the graph.

Abstract: The present invention relates generally to digital watermarking. In one implementation, we provide a hierarchical digital watermark detector method. The method includes: i) in a first layer of a hierarchical search, performing watermark detection on blocks of at least a portion of an incoming suspect signal; ii) identifying a first block in the portion that is likely to include a decodable digital watermark; and iii) in a second layer of the hierarchical search, performing additional watermark detection on overlapping blocks in a neighborhood around the first block. Another implementation provides a hierarchical watermark detector including a buffer and a detector. The buffer stores portions of an incoming signal. The detector evaluates watermark detection criteria for blocks stored in the buffer, and hierarchically zooms into a neighborhood of blocks around a block associated with watermark detection criteria that satisfies detection criteria.

Abstract: An image processing method is provided for an image processing apparatus which executes processing by allocating a plurality of weak discriminators to form a tree structure having branches corresponding to types of objects so as to detect objects included in image data. Each weak discriminator calculates a feature amount to be used in a calculation of an evaluation value of the image data, and discriminates whether or not the object is included in the image data by using the evaluation value. The weak discriminator allocated to a branch point in the tree structure further selects a branch destination using at least some of the feature amounts calculated by weak discriminators included in each branch destination.

Abstract: A method for fine-grained image classification on an image includes automatically segmenting one or more objects of interest prior to classification; and combining segmented and original image features before performing final classification.

Abstract: Using high-level attributes to guide image processing is described. In an embodiment high-level attributes of images of people such as height, torso orientation, body shape, gender are used to guide processing of the images for various tasks including but not limited to joint position detection, body part classification, medical image analysis and others. In various embodiments one or more random decision forests are trained using images where global variable values such as player height are known in addition to ground-truth data appropriate for the image processing task concerned. In some examples sequences of images are used where global variables are static or vary smoothly over the sequence. In some examples one or more trained random decision forests are used to find global variable values as well as output values for the task concerned such as joint positions or body part classes.

Abstract: A method for determining noise levels in a subband of an image. The method comprises receiving the subband of the image, defining block regions in at least two space domains of the subband, for each defined block region, identifying first wavelet coefficients associated with coordinate values in the at least two space domains in the defined block region, computing a correlation matrix between identified wavelet coefficients to determine the correlation between first wavelet coefficients according to the at least one color domain, computing second wavelet coefficients, the computation of second wavelet coefficients being based on the correlation matrix and the first wavelet coefficients, computing at least one noise level, the noise level computation being based on at least one second wavelet coefficient and providing the at least one noise level.

Abstract: A system and method are provided for model-based coronary artery calcium (CAC) scoring. A model image of a heart region is aligned with an image of a patient's heart region in order to more easily identify the coronary arteries and other relevant anatomical features in the image. Once the images are aligned, relevant calcium plaques are identified by their presence within a coronary artery, and the relevant plaques are then labeled by the specific coronary artery in which they are located. The coronary arteries with the labeled plaques are scored individually based on their size and X-ray attenuation, and an overall score based on all of the relevant plaques is then computed, which is related to the patient's risk for coronary artery disease.

Abstract: A speckle pattern is removed from each image such that each image obtained by a multi-resolution decomposition is decomposed into a low-band signal component, a high-band signal horizontal component, a high-band signal vertical component, and a high-band signal diagonal component, the decomposed signal components are subjected to a morphological reconstruction process, and then a multi-resolution reconstruction is performed.

Abstract: In one embodiment of the present invention, a graphics processing unit (GPU) is configured to detect an object in an image using a random forest classifier that includes multiple, identically structured decision trees. Notably, the application of each of the decision trees is independent of the application of the other decision trees. In operation, the GPU partitions the image into subsets of pixels, and associates an execution thread with each of the pixels in the subset of pixels. The GPU then causes each of the execution threads to apply the random forest classifier to the associated pixel, thereby determining a likelihood that the pixel corresponds to the object. Advantageously, such a distributed approach to object detection more fully leverages the parallel architecture of the PPU than conventional approaches. In particular, the PPU performs object detection more efficiently using the random forest classifier than using a cascaded classifier.

Abstract: Pattern matching of a plurality of stages that refer to respectively different parameters at each stage is performed with respect to each of a plurality of input data in sequence from a first stage until a matching result is false. A parameter referred to in pattern matching from a first stage until a predetermined stage among pattern matching of the plurality of stages is fixedly held in a fixed parameter holding unit. A parameter referred to in pattern matching of a stage after the predetermined stage is rewritably held in a variable parameter holding unit. In accordance with progress of pattern matching of the plurality of stages, a parameter held in the variable parameter holding unit is rewritten with an unheld parameter.

Abstract: A sort-out cycle such as a judgment of three levels of “usable”, “unusable”, and “reserve” is made to images and, when the judgment of all images is completed, the judgment of the three levels is made again to the “reserve” images is repeated. The number of sort-out cycle times in which the judgment is made is applied as an evaluation to the images which are finally determined as “usable”.

Abstract: Provided are an age estimation apparatus, an age estimation method, and an age estimation program capable of reducing the labor of labeling the image data used for age estimation. An age estimation apparatus for estimating an age of a person on image data includes a dimension compression unit for applying dimension compression to the image data to output low dimensional data; a clustering unit for performing clustering of the low dimensional data outputted; a labeling unit for labeling representative data of each cluster among the low dimensional data clustered; and an identification unit for estimating an age of a person on the basis of a learning result using a feature amount contained in labeled low dimensional data and unlabeled low dimensional data.

Abstract: In some embodiments, information associated with a first plurality of image pairs is received. Each image pair is assessed to detect visual pairwise differences that qualify as an error. A visual pairwise difference may be a difference with respect to at least one of position, size, color, or style. A prediction engine is trained based upon the assessed visual pairwise differences. Information associated with a second plurality of image pairs is received. Each of these image pairs comprises at least a portion of a visual end-user experience screen of an event-driven application executed in a client-tier environment. Each of these image pairs is assessed, using the prediction engine, to detect visual pairwise differences that qualify as an error. User feedback is received, indicating that at least one assessed pairwise difference should not have qualified as an error. The prediction engine is then re-trained based on the user feedback.

Abstract: Systems and methods are disclosed to categorize images by detecting local features for each image; applying a tree structure to index local features in the images; and extracting a rank list of candidate images with category tags based on a tree indexing structure to estimate a label of a query image.