85 results about "Re identification" patented technology

The present disclosure is directed to methods and systems for data-driven de-identification tool that can detect personally identifiable information (PII) elements within a given dataset and apply selective mapping transformations rules to de-identify the dataset. The disclosed data de-identification tool identifies direct identifiers, quasi-identifiers and unique values within a structured dataset in an example embodiment. The data de-identification tool then transforms these potentially personal or sensitive data elements into de-identified data elements and replaces the identified direct identifiers, quasi-identifiers and unique values within the structured dataset with the de-identified data elements. The data de-identification tool calculates a risk of re-identification and based on the risk level, repeat the de-identification process iteratively until the risk levels are within an acceptable range.

A method detects one or more occluded areas of a scene analysed by an object tracking system. The method includes building a map of one or more occluded areas in a scene. Building the map comprises running a re-identification algorithm on a video sequence to try to resume a lost object track. If the object track is successfully resumed, the method includes determining an area of the scene where the first object track is lost and an area of the scene where the first object track is resumed. A connection between the first and the second area of the scene is added the map such that the map identifies that an object track being lost in the first area of the scene has been resumed in the second area of the scene.

A system and method that allows privacy-enhanced querying to occur, where re-identification risk is reduced to a user-configured level are described. The system and method include a query federation agent that analyses and augments user-submitted queries to include results that contain metadata relating to the privacy characteristics. The system and method ensure results that contain privacy risks above defined thresholds are suppressed or altered so that re-identification cannot occur. The system and method add noise to results that contain privacy risks above defined thresholds so that re-identification cannot occur. The system and method utilize a data profile that defines the sources of potential re-identification risk in a database schema. The system and method apply privacy rules that are configurable for different database tables and configurable thresholds for different types of privacy risks.

A method is presented for detecting one or more occluded areas of a scene analysed by an object tracking system. The method includes building (S506) a map of one or more occluded areas in a scene. Building the map comprises running (5510) a re-identification algorithm on a video sequence to try to resume a lost object track. If the object track is successfully resumed, the method includes determining (S512) an area of the scene where the first object track is lost and an area of the scene where the first object track is resumed. A connection between the first and the second area of the scene is added (S514) the map such that the map identifies that an object track being lost in the first area of the scene has been resumed in the second area of the scene.

The invention discloses an online visual inspection device and method. Comprising the following steps: scanning a coil mold by using a surface structured light scanning camera, and identifying an initial point location; winding is carried out according to a coil winding technology, first scanning data in the winding process are obtained, and after winding of a first part of the coil is completed, the mold is turned; re-identifying the initial point position by adopting a surface structured light scanning camera, and winding the rest coil according to a coil winding process to obtain second scanning data; and according to the initial point location, the first scanning data and the second scanning data, calculating the coil width, segment spacing and long and short axes of each segment and each layer of coil to obtain a detection result. The measurement result of each section and each layer of wound coil can be obtained through on-line visual detection of the high-voltage coil, the structure is simple, the cost is low, detection is accurate, and full-automatic work is achieved.

A method for generating an optimised domain-generalisable model for re-identification of a target in a set of candidate images. The method optimises a local feature embedding model for domain-specific feature representation at each client of a plurality of clients, then receives, at a central server, information on changes to the local feature embedding model at each respective client resulting from the optimising step, and then updates a global feature embedding model based on the changes to the local feature embedding model. The method further receives, at each client from the central server, information representative of the updates to the global feature embedding model, then maps, at each client, on to the respective local feature embedding model at least a portion of the received updates, and subsequently updates, at each client, the respective local feature embedding model based on the mapped updates. The steps are repeated until convergence criteria are met, wherein the global feature embedding model is the optimised domain-generalisable model for re-identification of a target in a set of candidate images.

The invention discloses an ancient building construction drawing automatic verification method based on machine learning, and the method comprises the following steps: carrying out the preprocessing of a construction drawing image, and extracting clear boundary information; secondly, extracting geometric structure features of the primitives through an image analysis technology, such as line segments, curves and closed regions; secondly, generating a classification template by utilizing historical drawing data, and establishing an identification model by combining geometric features and semantic information of the primitives; the model can intelligently classify and accurately identify each component in a drawing, and judges whether the structure is matched with the semantics or not. When a suspected abnormal area is found, the system automatically recognizes and marks the suspected abnormal area again, a set of complete drawing verification process is formed, and the method has universality suitable for various ancient building style drawings. The invention aims to reduce the time and error rate of manual drawing auditing.

Systems and methods for tracking and re-identifying tracked subjects in an area of real space are disclosed. The method includes generating first and second reidentification feature vectors of a first subject identified from a first time interval from the respective first and second sequences of images as obtained from the first time interval. The method includes generating third and fourth reidentification feature vectors of a second subject identified from a second time interval from the respective first and second sequences of images. The method includes matching a second subject identified from a second time interval with the first subject identified from the first time interval when at least one of a first similarity score between the first and the third reidentification feature vectors and a second similarity score between the second and the fourth reidentification feature vectors is above a pre-defined threshold.

The method delivers a degree of protection (txP3) representative of the risk of re-identification of data in the case of a correspondence search attack including a deterministic search based on an external information source and a correspondence search based on a distance. The method comprises steps of E) consolidating a set of original individuals (EDO) and a set of anonymous individuals (IA); F) identifying, in the set of original individuals, individuals at risk (IOrs) via the deterministic correspondence search; G) evaluating a degree of failure of re-identification (txP1) for the sets of original individuals and of anonymous individuals, on the basis of the correspondence search based on distance; H) computing the degree of protection as a function of a total number of individuals in the original dataset, of a number (RS) of individuals at risk identified in step B) and of the degree of failure of re-identification (txP1).

The present invention relates to the technical field of energy storage management, and discloses an energy storage management system based on an AI recognition model; the method comprises acquiring communication characteristics of each subunit, identifying an adapter, constructing an energy storage management scenario, acquiring comprehensive management parameters of the energy storage management scenario, calculating a management reliability index of the energy storage management scenario, determining whether to generate a re-identification prompt, formulating a re-identification instruction, and re-identifying the adapter; compared with the prior art, the present invention can specifically predict an adapter corresponding to a communication protocol and a point table through an AI recognition model, and can accurately calculate the adaptation operation reliability between the identified adapter and the subunit, and can re-identify the adapter when it is found that the degree of adaptation is not good, thereby ensuring that a fast and accurate data link and control chain conduction adaptation effect can be achieved between the subunits of the energy storage management system, thereby improving the operation efficiency of the energy storage management system.

A data anonymization pipeline system for managing holding and pooling data is disclosed. The data anonymization pipeline system transforms personal data at a source and then stores the transformed data in a safe environment. Furthermore, a re-identification risk assessment is performed before providing access to a user to fetch the de-identified data for secondary purposes.

The invention discloses a vehicle multi-target tracking method capable of keeping a unique mark number, and belongs to the field of computer vision, aiming at the problem that the target mark number is not unique after the observation angle is changed or shielded during multi-target tracking. Vehicle videos blocked at the same observation angle or at different observation angles are collected to construct a data set, and the data set is processed according to the following process: inputting a video 1, tracking to obtain a vehicle target label and position, and intercepting a target picture frame by frame to form the data set for training a DenseNet121 network; inputting a video 2, tracking to obtain target information, inputting a target picture into the trained DenseNet121 network for re-identification, reserving an old label for an old target, endowing a new label for a new target, intercepting a new target picture to expand a data set, and updating the training network by using the data sets of the video 1 and the video 2; and inputting a video 3 and a subsequent video, and repeating the steps, namely tracking the target, re-identifying the label, expanding the data set and updating network training.

It is possible to appropriately set a threshold for detecting anomalies. [Solution] A threshold specification unit 22 specifies a threshold value using a target value in extracted data extracted from performance data. An abnormality detection unit 23 uses the threshold value to detect abnormal data in the extracted data, in which the target value is an abnormal value. A designation receiving unit 24 indicates the abnormal data and receives designation of outlier data in which the target value is an outlier among the abnormal data. The threshold specification unit 22 respecifies the threshold value using a target value in the remaining data obtained by excluding the outlier data from the extracted data. The abnormality detection unit 23 uses the respecified threshold value to detect abnormal data in which the target value is an abnormal value among the remaining data.

This invention relates to a panoramic target search and monocular linkage tracking method, apparatus, and computer device. The method includes acquiring target information; identifying the target in a panoramic image using a target detection algorithm; determining the target monocular camera corresponding to the target from the panoramic image; tracking the target using the target monocular camera according to a preset method, the preset method including cyclically switching between a monocular tracking mode of a first preset duration and a panoramic shooting mode of a second preset duration; when the target monocular camera switches to the panoramic shooting mode, re-synthesizing an updated panoramic image; determining whether the target has been lost from the field of view of the target monocular camera in monocular tracking mode; if so, re-identifying the target from the updated panoramic image. The linkage mechanism between monocular and panoramic tracking in this invention allows for flexible adjustment of the tracking strategy, maintaining tracking accuracy and stability, and enhancing tracking robustness in complex scenes.

The invention relates to a method and system for re-identification. The method includes: processing first original image data by applying a first image processing setting; detecting a first object; extracting a first feature vector; processing the second original image data by applying a second image processing setting; detecting a second object; extracting a second feature vector; determining the similarity of the first object and the second object by comparing the first feature vector with the second feature vector; if the difference between the first feature vector and the second feature vector exceeds a first threshold and the difference between the first image processing setting and the second image processing setting exceeds a second threshold, reprocessing the first original image data corresponding to the first image region by applying the second image processing setting; extracting an updated first feature vector from the first image region; the updated first feature vector is compared to the second feature vector. The disclosure further relates to an image processing system configured to perform the method.

The present disclosure is directed to methods and systems for data-driven de-identification tool that can detect personally identifiable information (PII) elements within a given dataset and apply selective mapping transformations rules to de-identify the dataset. The disclosed data de-identification tool identifies direct identifiers, quasi-identifiers and unique values within a structured dataset in an example embodiment. The data de-identification tool then transforms these potentially personal or sensitive data elements into de-identified data elements and replaces the identified direct identifiers, quasi-identifiers and unique values within the structured dataset with the de-identified data elements. The data de-identification tool calculates a risk of re-identification and based on the risk level, repeat the de-identification process iteratively until the risk levels are within an acceptable range.

The invention discloses a vehicle re-identification domain adaptive method based on a visual language model, and belongs to the field of image processing and machine learning. The problems that in the prior art, semantic information obtaining means are lacked, and the re-recognition accuracy is low are solved. The visual text representation capability of the visual text model is fully utilized, and the cross-modal description capability of the visual language model is utilized by distributing a group of learnable text marks for each individual. These markers are fed to a text encoder to generate cues, enhancing feature extraction of the target domain by the image encoder. In the constraint prompt learning step, an image encoder and a text encoder are fixed, and only the text prompt is optimized. In the joint image encoder optimization stage, source domain data and target domain data are used in a joint mode, text prompt and a text encoder are fixed, and constraints are provided for fine adjustment of the image encoder. According to the method, the feature extraction capability of the model is enhanced in a vehicle re-identification domain adaptive task, and the matching accuracy is improved.

A method of object re-identification is disclosed, in particular a method of object re-identification in images of objects. The method comprises providing a plurality of neural networks (27) for object re-identification, wherein each of the plurality of neural networks is trained on image data of a different group of anatomical features, each group being represented by a reference vector; receiving a plurality of images (4) of an object (38) and an input vector representing anatomical features depicted in all of the plurality of images (4); comparing the input vector to the reference vectors to determine a most similar reference vector according to a predefined condition; and inputting image data of the plurality of objects (38) to the neural network (#1) represented by the most similar reference vector to determine whether the plurality of objects (38) have the same identity.

The present application relates to the field of personnel re-tracking, and particularly relates to a personnel re-tracking method based on image detection technology. The method first acquires a monitoring image of a monitoring area in real time, performs human body detection on a current frame image, screens out a tracking target in the current frame image, performs skeleton point detection on the tracking target in the current frame image, and judges whether the tracking target is a human body target according to the number of skeleton points, generates a distance relationship table containing the distance between skeleton points of the human body target for the current frame image, judges whether the human body target is lost according to the number of images containing the human body target, and re-identifies and tracks the human body target in a new monitoring image according to the difference between a new distance relationship table and a historical distance relationship table. The present application can solve the problems of high target false detection rate in the existing personnel tracking algorithm, low re-identification accuracy after tracking loss and weak anti-interference ability.

Controlling a list of inactive trajectories for re-identification in an object tracking system is disclosed. The present disclosure relates to methods (600), systems, and software for controlling a list of inactive trajectories in an object tracking system. The described techniques include using a decision threshold to obtain (S604) a false alarm rate (FPR) of a re-identification model and obtain (S606) an acceptable FPR of a tracking system. The re-identification model is used to match the new detection with the inactive trajectory. The technique includes determining (S608) one or more termination conditions for deleting the inactive trajectory based on the acceptable FPR and the model FPR. If the first inactivity trajectory satisfies a termination condition, the first inactivity is deleted from the list (S610).

A method, system and software for controlling a list of inactive tracks in an object tracking system. The techniques described includes obtaining the false positive rate, FPR, of a re-identification model using a decision threshold, and obtaining an acceptable FPR of the tracking system. The re-identification model is used to match new detections with inactive tracks. The techniques include determining one or more terminal conditions for deleting inactive tracks based on the acceptable and model FPRs. If a first inactive track meets a terminal condition, it is deleted from the list.

In an embodiment an apparatus includes a processor configured to generate a feature extraction module using a dataset in which an attribute for each object is defined, receive an image obtained by a camera, extract an attribute of an object of interest from the image using the learned feature extraction module, identify an object re-identification candidate group based on the extracted attribute of the object of interest and re-identify the object of interest based on the identified object re-identification candidate group.

A system for monitoring safety compliance comprises a plurality of cameras and a computing system. The plurality of cameras are configured to obtain data, and the data comprises multiple images associated with one or more objects. The computing system is configured to process the data to determine the safety compliance of the one or more objects associated with the multiple images based on implementing a first model trained for re-identification. The computing system is further configured to train the first model for re-identification, by determining a similarity loss and updating the first model based on the similarity loss.

A computer-implemented method of tracking objects in a video sequence of a scene comprises determining a location of a sink in the scene where objects exit the scene and a location of a source where objects enter the scene; tracking a first object moving in the scene using a re-identification algorithm, wherein the first object is associated with a re-identification threshold of the re-identification algorithm; detecting that the first object has exited the scene at the sink; and responsive to detecting that the first object has exited the scene at the sink, adjusting the re-identification threshold associated with the first object such that a probability that the re-identification algorithm re-identifies a second object, entering the scene at the source after the first object has exited the scene at the sink, as the first object is reduced.

A method for generating an optimised domain-generalisable model for re-identification of a target in a set of candidate images. The method optimises a local feature embedding model for domain-specific feature representation at each client of a plurality of clients, then receives, at a central server, information on changes to the local feature embedding model at each respective client resulting from the optimising step, and then updates a global feature embedding model based on the changes to the local feature embedding model. The method further receives, at each client from the central server, information representative of the updates to the global feature embedding model, then maps, at each client, on to the respective local feature embedding model at least a portion of the received updates, and subsequently updates, at each client, the respective local feature embedding model based on the mapped updates. The steps are repeated until convergence criteria are met, wherein the global feature embedding model is the optimised domain-generalisable model for re-identification of a target in a set of candidate images.

An example apparatus for re-identifying objects includes an image receiver to receive a first image and a second image of an object with an identity. The apparatus also includes a fused model generator to fuse a global representation of the object with local representations of pose parts of the object to generate a fused representation of the object based on the first image. The apparatus further includes an object re-identifier to re-identify the object with the identity in the second image based on the fused representation.

According to the human re-identification method according to the present disclosure, first, estimating a pose of a human to be re-identified in an image of the human is executed. Next, clipping a predetermined number of patches from the image along a body of the human based on the pose is executed. Then, generating positional information of each of the predetermined number of patches is executed. Then, inputting the predetermined number of patches with the positional information into a vision transformer encoder is executed. Then, inputting an output of the vision transformer encoder into a neural network is executed. And finally, obtaining an output of the neural network as a re-identification result of the human is executed.

A system and method for distributed data intersection using privacy preserving collaborative detection is disclosed The system includes a processor and a memory that causes the processor to verify data sets and configurations from a first and second entity interacting in a distributed data intersection to ensure data integrity and privacy during a data collaboration. The processor enables computations between the first and second entity using cryptographic protocols to identify common data points. The processor allows the first and second entity to retrieve the common data points pertaining to other entity using a oblivious data sharing technique while maintaining confidentiality of the data accessed during the interaction. The processor facilitates validation of data integrity and accuracy without revealing underlying information. The processor controls data-sharing policies ensuring that authorized data is disclosed based on a predefined criteria thereby enabling collaboration of data for re-identification while protecting privacy.

A system for providing data privacy risk of data privacy related data objects for assigning an access right for processing and storing the data objects, including receiving more than one data object and extracting data from the data objects, generating at least one identity space, wherein the identity space includes those data objects which include extracted data elements concerning a group of individuals or a subset of the group of individuals, and wherein the group of individuals differs from the group of individuals of each other identity space, calculating at least one privacy risk for each identity space and for the contained individuals depending on re-identification probabilities of each individual based on the extracted data elements in the identity space and depending on a set of background knowledge which is assumed to be known, exporting the extracted data and export the calculated privacy risk.