13 results about "Back tracking" patented technology

Disclosed are apparatuses, systems, and techniques for implementing bidirectional tracking in computer vision applications. In one embodiment, the techniques include obtaining digital representations of an object depicted in video frames and for each of a forward direction (FD) of tracking and a reverse direction (RD) of tracking, obtaining, using (i) a current state of the object associated with an upstream video frame and (ii) the digital representation of the object for a downstream video frame, an updated state of the object associated with the downstream video frame. The techniques further include obtaining, using the updated state of the object for the FD and / or the updated state of the object for the RD, a bidirectional state of the object, and determining, using the bidirectional state of the object, a trajectory of the object across the video frames.

A system configured to determine poses of a tracked device in a physical environment and to utilize the poses as an input to control or manipulate a virtual environment or mixed reality environment. In some cases, the system may include a fall back tracking system for when the main tracking system loses visual tracking of the tracked device.

The present invention discloses an unsupervised RGB-T object tracking method based on attention multi-modal feature fusion. First, a hierarchical convolutional neural network is used to extract the features of RGB images and thermal infrared images. Then, a feature fusion module is used to synchronously fuse the features from different levels and different modalities. Next, the fused features are subjected to two forward tracking operations to obtain a response map. Then, the fused features are reversed, the original template image is used as the search image, the search image is used as the template image, and the generated response map is used as a pseudo-label for backward tracking to obtain the final response map. Then, the consistency loss between the response map obtained by backward tracking and the original label is minimized for unsupervised training. Finally, the test video frames are input into the trained network for forward tracking to obtain the response map, which is the predicted target position. The method of the present invention can make full use of multi-level and multi-modal information and can give play to the advantages of unsupervised learning.

A Viterbi traceback processing method, system, and apparatus is provided in which a first Viterbi traceback processing operation (MUX 514) is performed on a first survivor path metric (TMV1) by selecting a first output data bit (Ti1) for the first survivor path metric in response to a reverse tracking state (INDEX 0), in which a plurality of Viterbi traceback processing operations (MUX 512, 513) are performed on respective portions of additional survivor path metrics (TMV2A, TMV2B) by selecting candidate data bits (Tn1, Tn2) for the additional survivor path metrics in response to shifted reverse tracking states (INDEX 1), in which a multiplexer (MUX 518) controlled by the first output data bit selects between the candidate data bits to generate an additional output data bit (Ti2) for the additional survivor path metrics, such that Viterbi traceback processing operations are performed in parallel to produce output data bits.

The invention discloses a robot vision simulation inspection system and method, and relates to the technical field of intelligent inspection. The system comprises an environment monitoring module which is used for circumferentially moving around an explosive reservoir area, rotating a monitoring field angle in the moving process, and collecting environment parameters; the fusion visual motion algorithm module is used for acquiring the position, attribute parameters and state parameters of the intrusion target according to the environmental parameters; the dynamic and static target monitoring module is used for moving around an explosive reservoir area and rotating a field angle according to the position of the invasion target, tracking the invasion target, and feeding back tracking information and attribute parameters and state parameters of the invasion target to a monitoring room; and the dynamic regulation and control module is used for regulating the moving speed of the environment monitoring module and the rotating speed of a monitoring field angle according to the environment parameters, and regulating the moving speed and the angle of the field angle of the dynamic and static target monitoring module according to the multi-dimensional information of the invasion target. Through the system, the potential safety hazard in the explosive reservoir area can be accurately detected.

The invention discloses a target rematching method based on a motion trail in single target tracking, and relates to the technical field of intelligent visual monitoring and video analysis. The method comprises the following steps: firstly, screening a target candidate box, and performing multi-dimensional punishment and screening on a current frame tracking result to obtain the target candidate box; then, whether tracking ambiguity exists or not is judged in combination with a historical real trajectory; and then constructing a two-dimensional motion model based on Kalman filtering, and predicting the current position of the target. And finally, constructing a trajectory pool through reverse tracking, solving bipartite graph maximum matching by using a Hopcroft-Karp algorithm to realize target re-matching, and introducing an interference target trajectory pool dynamic updating mechanism to improve the matching accuracy. According to the method, tracking deviation caused by similar target interference can be effectively corrected, long-term stable tracking of an initial target by a tracker is ensured, the method is suitable for various single-target tracking scenes with similar target interference, and the long-term tracking robustness in complex scenes is remarkably improved.

Arrangements for providing semi-dynamic vulnerability detection are provided. In some aspects, a computing platform may receive work flow data from one or more systems and may analyze the work flow data using a GAN. The GAN may output a potential vulnerability identified in the data, and a category of the potential vulnerability. Based on the potential vulnerability and the category, the computing platform may determine a severity of the potential vulnerability. An ANN-SNN converter may be executed to output a knowledge graph including a plurality of nodes forming a mitigation action plan for the potential vulnerability. The computing platform may generate a digital twin of the knowledge graph and may then reconcile the digital twin by back tracking through each node to validate each node of the digital twin. Based on the digital twin being reconciled, the generated mitigation action plan may be transmitted to a computing system for execution.

The invention discloses a power equipment data processing method and system based on an optical fiber sensing technology, and relates to the technical field of power equipment monitoring, and the method comprises the steps: extracting abnormal change characteristics from a virtual reference correction table, carrying out the interpretive discrimination through combining with the state of power equipment, respectively writing in an enhancement candidate and a suppression candidate, and generating a double-mark candidate book; performing time undertaking check, spatial adjacency check and parallel arbitration on the double-label candidate book, dynamically distinguishing dominant abnormal entries and standby abnormal entries, and generating a main and standby abnormal hitching list; and performing hierarchical packaging, instruction assignment and trace collection on the main / standby abnormal hooking list, outputting real-time monitoring content, synchronously writing back a tracking processing instruction and a trace record, and generating an optical fiber data processing output set. According to the invention, space-time check and parallel arbitration are realized, and the monitoring accuracy of the power equipment is improved.

The application discloses a kind of self-supervised tracking loss optimization methods based on multi-cycle contrast enhancement consistency, belong to target tracking technical field.The method is: the video frame in training data set is successively executed forward tracking and reverse tracking, and data enhancement operation is carried out to template frame in last step of forward tracking and reverse tracking;According to the real target label of first frame and the pseudo label generated by its reverse tracking, cycle consistency loss is constructed;According to the difference between enhanced video frame and corresponding original video frame in executing forward tracking and reverse tracking, contrast enhancement loss is constructed;Self-supervised tracking model is optimized by combining cycle consistency loss and contrast enhancement consistency loss.The application simulates target appearance change using data enhancement technology, constructs contrast enhancement consistency loss in video frame to optimize tracking model, and realizes high-performance tracking.

Disclosed are methods for authorizing access to reverse tracking data related to biodegradability of materials by a material producer, the method comprising the steps of: providing at least one authorization rule associated with the material, wherein the at least one authorization rule provides access for the material producer to material data generated in relation to a production, use and / or an end-of-life stage of the product produced using the material, generating tracking data associated with at least one property related to the biodegradability of the material in the production, use and / or an end-of-life stage by applying the at least one authorization rule to the material data, and providing the tracking data to a data providing service of the product producer and / or product user for access by a data consuming service of the material producer.

The invention discloses an air-ground infrared target tracking method based on consistency reasoning correlation filtering, and relates to the field of aircraft visual target tracking, and the method comprises the steps: determining a target search region of a current frame according to the target position of a previous video frame, predicting the target state of the current video frame through a consistency reasoning correlation filter, and obtaining the target state of the current video frame; obtaining a target response of the current video frame; determining a target position of the current video frame based on the obtained target response, and determining a target size by adopting a scale correlation filter; determining the target response reliability based on the target response of the current video frame, determining a dynamic threshold by combining the target response of the historical video frame, and when the target response reliability is greater than the dynamic threshold, updating a consistency reasoning correlation filter by adopting front and back tracking consistency regularization and filter time regularization, and target tracking of the next video frame is carried out. According to the invention, the method can achieve the target tracking in a real-time and robust manner while improving the discrimination capability of a correlation filter.

The present application relates to a vulnerability testing method, device, computer equipment and storage medium. The method obtains the source code to be tested, and cuts the source code according to preset rules to obtain a token file, and then uses an analysis engine to track the token file to determine the input interface of the source code, and then performs a backtracking check of the variable tracking process by forward and reverse tracing the variables associated with the input interface of the source code, finds vulnerabilities in the transmission process, and performs semantic analysis and grammatical analysis on the code to obtain a syntax tree; traverses the syntax tree and combines the vulnerabilities in the variable transmission process found by forward and reverse tracing the variables associated with the input interface of each code, and determines the logical relationship between the network address path corresponding to the vulnerability after the source code is compiled by the system and the vulnerability position in the source code. Better display the relationship between the source code and the system code level, better serve testers to reproduce and retest existing vulnerabilities, and effectively improve the code quality of programmers.

The invention relates to the technical field of intelligent traffic, in particular to a vehicle cross-boundary tracking method and device, electronic equipment and a storage medium, the method is applied to a vehicle tracking system, and the vehicle tracking system comprises a management platform and an algorithm platform which are connected in sequence; the method comprises the following steps: through a configuration rule line of a management platform, establishing a logic pairing relationship between an entry rule line and an exit rule line, so as to generate an algorithm task containing a camera ID, a rule line coordinate and a region pairing relationship; and the algorithm platform receives the algorithm task, performs vehicle cross-border tracking based on configuration data of the algorithm task, and feeds back a tracking result to the management platform. The monitoring rule is configured based on the management platform, the monitoring platform is controlled to perform vehicle cross-border tracking based on the monitoring rule under the action of the task platform, vehicle cross-border tracking in a complex scene can be adapted, the track integrity is improved, and the error rate is reduced.