The present invention relates to a character information recognition method based on image processing. The method comprises: collecting images to obtain a target character image; then sequentially comparing the target character image with character template images in a character template library to find a maximum of a coincidence area of the character in the target character image with the character templates in the character template images; and when the coincidence area meets a preset condition, determining the target character to be recognized as the character in the corresponding character template image. The character templates are designed to include not only a coincidence-permitted region but also a coincidence-restricted region. The coincidence-restricted region is set, so that the direct comparing and matching of the character templates can be more accurately carried out, thereby improving the recognition speed.

One or more image parameters of an image may be analyzed using a hierarchical set of models. Executing individual models in the set of models may generate outputs from analysis of different image parameters of the image. Inputs of one or more of the models may be conditioned on a set of outputs derived from one or more preceding model in the hierarchy.

Systems and methods for censoring text-based data are provided. In some embodiments a censoring system may include at least one processor and at least one non-transitory memory storing application programming interface instructions. The censoring system may be configured to perform operations comprising storing a target pattern type and a computer-based model for identifying a target data pattern corresponding to a target pattern type within text based data. The censoring system may also be configured to receive text-based data by a server, and to retrieve the stored target pattern type to be censored in the text-based data. The censoring system may be configured to identify within the received text-based data, a target data pattern corresponding to the retrieved target pattern type. The censoring system may be configured to censor target characters within the identified target data pattern, and transmit the censored text-based data to a receiving party.

A method for testing medical data is provided. Each medical datum includes a plurality of information units and a plurality of separators, and the method includes the following steps: a. matching the medical data against a standard library including a plurality of patterns, a matching expression being:

[\s\S][number/sequence/relation]&[\b|\B] (S101); and b. determining, based on a matching result of the step a, whether the medical datum is qualified (S102). A standardized standard library is first established, a matching result is obtained by matching the medical datum and the standard library for a non-initial boundary, an initial boundary, an information quantity, information sequences, a semantic relationship quantity, a character boundary, and a non-character boundary, and whether the medical datum meets a requirement is further determined according to the matching result.

Examples relate to simultaneous localization and calibration. An example implementation may involve receiving sensor data indicative of markers detected by a sensor on a vehicle located at vehicle poses within an environment, and determining a pose graph representing the vehicle poses and the markers. For instance, the pose graph may include edges associated with a cost function representing a distance measurement between matching marker detections at different vehicle poses. The distance measurement may incorporate the different vehicle poses and a sensor pose on the vehicle. The implementation may further involve determining a sensor pose transform representing the sensor pose on the vehicle that optimizes the cost function associated with the edges in the pose graph, and providing the sensor pose transform. In further examples, motion model parameters of the vehicle may be optimized as part of a graph-based system as well or instead of sensor calibration.

Systems and methods for generating synthetic video are disclosed. For example, a system may include a memory unit and a processor configured to execute the instructions to perform operations. The operations may include receiving video data, normalizing image frames, generating difference images, and generating an image sequence generator model. The operations may include training an autoencoder model using difference images, the autoencoder comprising an encoder model and a decoder model. The operations may include identifying a seed image frame and generating a seed difference image from the seed image frame. The operations may include generating, by the image sequence generator model, synthetic difference images based on the seed difference image. In some aspects, the operations may include using the decoder model to synthetic normalized image frames from the synthetic difference images. The operations may include generating synthetic video by adding background to the synthetic normalized image frames.

An image collating device that collates a first image and a second image includes a frequency characteristic acquiring unit, a frequency characteristic synthesizing unit, and a determining unit. The frequency characteristic acquiring unit acquires a frequency characteristic of the first image and a frequency characteristic of the second image. The frequency characteristic synthesizing unit generates a synthesized frequency characteristic by synthesizing the frequency characteristic of the first image and the frequency characteristic of the second image. The determining unit calculates a score indicating a degree to which the synthesized frequency characteristic is a wave having a single period, and collates the first image and the second image based on the score.

Systems and methods for efficiently training a machine learning model are presented. More particularly, using information regarding the relevant neighborhoods of target nodes within a body of training data, the training data can be organized such that the initial state of the training data is relatively easy for a machine learning model to differentiate. Once trained on the initial training data, the training data is then updated such that differentiating between a matching and a non-matching node is more difficult. Indeed, by iteratively updating the difficulty of the training data and then training the machine learning model on the updated training data, the speed that the machine learning model reaches a desired level of accuracy is significantly improved, resulting in reduced time and effort in training the machine learning model.

Methods for generating data models using a generative adversarial network can begin by receiving a data model generation request by a model optimizer from an interface. The model optimizer can provision computing resources with a data model. As a further step, a synthetic dataset for training the data model can be generated using a generative network of a generative adversarial network, the generative network trained to generate output data differing at least a predetermined amount from a reference dataset according to a similarity metric. The computing resources can train the data model using the synthetic dataset. The model optimizer can evaluate performance criteria of the data model and, based on the evaluation of the performance criteria of the data model, store the data model and metadata of the data model in a model storage. The data model can then be used to process production data.

A character recognition method, a character recognition apparatus, an electronic device and a computer readable storage medium are disclosed. The character recognition method includes: determining semantic information and first position information of each individual character recognized from an image; constructing a graph network according to the semantic information and the first position information of each individual character; and determining a character recognition result of the image according to a feature of each individual character calculated by the graph network.