Character recognition processing suitable to a handwritten character area and a printed character area among character areas in a scanned image of a document is performed. Next, character recognition results for the handwritten character area and character recognition results for the printed character area are integrated and a likelihood indicating a probability of being an extraction target is calculated for a candidate character string that is an extraction candidate among the integrated character recognition results and a character string that is the item value is determined. Then, at the time of the determination, different evaluation indications are used in a case where a character originating from the handwritten character area is included in characters constituting the candidate character string and in a case where such a character is not included.

Aspects of the present invention provide methods, apparatuses, systems, computing devices, computing entities, and/or the like for identifying data processing activities associated with various data assets based on data discovery results. In accordance various aspects, a method is provided comprising: identifying and scanning data assets to detect a subset of the data assets, wherein each asset of the subset is associated with a particular data element used for target data; generating a prediction for each pair of data assets of the subset on the target data flowing between the pair; identifying a data flow for the target data based on the prediction generated for each pair; and identifying a data processing activity associated with handling the target data based on a correlation identified for the particular data element, the subset, and/or the data flow with a known data element, subset, and/or data flow for the data processing activity.

Embodiments include a method, an electronic device, and a computer program product for information processing. In an example embodiment, a method for information processing includes: acquiring, at a first device, a first feature associated with a target object; applying the first feature to a trained first model deployed at the first device to determine a first confidence coefficient, the first confidence coefficient being associated with probabilities that the first model determines the target object as a real object and as a false object; if the first confidence coefficient is lower than a first threshold confidence coefficient, sending a request for verifying the target object to a second device, the second device being deployed with a trained second model for verifying the target object, and the second model being more complex than the first model; and updating the first model based on a response to the request.

An example apparatus for mining multi-scale hard examples includes a convolutional neural network to receive a mini-batch of sample candidates and generate basic feature maps. The apparatus also includes a feature extractor and combiner to generate concatenated feature maps based on the basic feature maps and extract the concatenated feature maps for each of a plurality of received candidate boxes. The apparatus further includes a sample scorer and miner to score the candidate samples with multi-task loss scores and select candidate samples with multi-task loss scores exceeding a threshold score.

The present application provides an image feature acquisition method and a corresponding apparatus. According to an example of the method, a classification model may be trained by using preset classes of training images, and similar image pairs may be determined based on the training images; classification results from the classification model are tested by using verification images to determine nonsimilar image pairs; and the classification model is optimized based on the similar image pairs and the nonsimilar image pairs. In this way, the optimized classification model may be used to acquire image features.

A system for image segmentation is provided. The system may obtain a target image including an ROI, and segment a preliminary region representative of the ROI from the target image using a first ROI segmentation model corresponding to a first image resolution. The system may segment a target region representative of the ROI from the preliminary region using a second ROI segmentation model corresponding to a second image resolution. At least one model of the first and second ROI segmentation models may at least include a first convolutional layer and a second convolutional layer downstream to the first convolutional layer. A count of input channels of the first convolutional layer may be greater than a count of output channels of the first convolutional layer, and a count of input channels of the second convolutional layer may be smaller than a count of output channels of the second convolutional layer.

A method for retrieving an image is provided. The method includes: extracting a global feature and a local feature of an image to be retrieved, and a global feature and a local feature of an image to be recalled by employing a preset neural network model; determining a candidate image set by matching the global feature of the image to be retrieved with the global feature of the image to be recalled and matching the local feature of the image to be retrieved with the local feature of the image to be recalled; and determining a retrieval result from the candidate image set by performing local feature verification on the image to be retrieved and a candidate image in the candidate image set. An apparatus for retrieving an image, an electronic device, and a medium are further provided.

A computer implemented method for optical character recognition (OCR) of a character string in a text image. The method efficiently combines two different OCR engines with the computation that needs to be done by the second OCR engine depending on the results found by the first OCR engine. This method provides, in particular, a high speed and accurate results when the first OCR engine is fast and the second OCR engine is accurate. The combination is possible because the second OCR engine identifies each segment to be processed by the second OCR engine without needing to process all segments.

Aspects of the subject matter disclosed herein include methods, systems, and other techniques for training, in a first phase, an object classifier neural network with a first set of training data, the first set of training data including a first plurality of training examples, each training example in the first set of training data being labeled with a coarse-object classification; and training, in a second phase after completion of the first phase, the object classifier neural network with a second set of training data, the second set of training data including a second plurality of training examples, each training example in the second set of training data being labeled with a fine-object classification.

A computer-implemented method for determining scene-consistent motion forecasts from sensor data can include obtaining scene data including one or more actor features. The computer-implemented method can include providing the scene data to a latent prior model, the latent prior model configured to generate scene latent data in response to receipt of scene data, the scene latent data including one or more latent variables. The computer-implemented method can include obtaining the scene latent data from the latent prior model. The computer-implemented method can include sampling latent sample data from the scene latent data. The computer-implemented method can include providing the latent sample data to a decoder model, the decoder model configured to decode the latent sample data into a motion forecast including one or more predicted trajectories of the one or more actor features. The computer-implemented method can include receiving the motion forecast including one or more predicted trajectories of the one or more actor features from the decoder model.