A domain alignment technique for cross-domain object detection tasks is introduced. During a preliminary pretraining phase, an object detection model is pretrained to detect objects in images associated with a source domain using a source dataset of images associated with the source domain. After completing the pretraining phase, a domain adaptation phase is performed using the source dataset and a target dataset to adapt the pretrained object detection model to detect objects in images associated with the target domain. The domain adaptation phase may involve the use of various domain alignment modules that, for example, perform multi-scale pixel/path alignment based on input feature maps or perform instance-level alignment based on input region proposals.

The present disclosure provides a training method of a text recognition model, a text recognition method, and an apparatus, relating to the technical field of artificial intelligence, and specifically, to the technical field of deep learning and computer vision, which can be applied in scenarios such as optional character recognition, etc. The specific implementation solution is: performing mask prediction on visual features of an acquired sample image, to obtain a predicted visual feature; performing mask prediction on semantic features of acquired sample text, to obtain a predicted semantic feature, where the sample image includes text; determining a first loss value of the text of the sample image according to the predicted visual feature; determining a second loss value of the sample text according to the predicted semantic feature; training, according to the first loss value and the second loss value, to obtain the text recognition model.

A mechanism is described for facilitating embedding of human labeler influences in machine learning interfaces in computing environments, according to one embodiment. A method of embodiments, as described herein, includes detecting sensor data via one or more sensors of a computing device, and accessing human labeler data at one or more databases coupled to the computing device. The method may further include evaluating relevance between the sensor data and the human labeler data, where the relevance identifies meaning of the sensor data based on human behavior corresponding to the human labeler data, and associating, based on the relevance, human labeler data with the sensor data to classify the sensor data as labeled data. The method may further include training, based on the labeled data, a machine learning model to extract human influences from the labeled data, and embed one or more of the human influences in one or more environments representing one or more physical scenarios involving one or more humans.

A model maintenance device according to an embodiment performs maintenance of a model for pattern recognition used in label estimation of target data for recognition. The model maintenance device includes a generating unit, an evaluating unit, and an updating determining unit. The generating unit generates a new model using learning data. The evaluating unit evaluates the performance of the new model using evaluation data classified into a first group, from among evaluation data classified into a plurality of groups, and calculates a first performance evaluation value; and evaluates performance of the new model using evaluation data classified into a second group, from among the evaluation data classified into a plurality of groups, and calculates a second performance evaluation value. Based on the first performance evaluation value and the second performance evaluation value, the updating determining unit determines whether or not the existing model should be updated with the new model.

A method and apparatus for building an image model, where the apparatus generates a target image model that includes layers duplicated from a layers of a reference image model and an additional layer, and trains the additional layer.

In some embodiments, reduction of computational resource usage related to image labeling and/or segmentation may be facilitated. In some embodiments, a collection of images may be used to train one or more prediction models. Based on a presentation of an image on a user interface, an indication of a target quantity of superpixels for the image may be obtained. The image may be provided to a first prediction model to cause the prediction model to predict a quantity of superpixels for the image. The target quantity of superpixels may be provided to the first model to update the first model's configurations based on (i) the predicted quantity and (ii) the target quantity. A set of superpixels may be generated for the image based on the target quantity, and segmentation information related to the superpixels set may be provided to a second prediction model to update the second model's configurations.

An information processing method includes: reading a layer structure and parameters of layers from each of models of two neural networks; and determining a degree of matching between the models of the two neural networks, by comparing layers, of the respective models of the two neural networks, that are configured as a graph-like form in respective hidden layers, in order from an input layer using breadth first search or depth first search, based on similarities between respective layers.

System and method of deploying a trained machine learning neural network (MLNN) in generating a fall injury condition of a subject. The method comprises receiving, at input layers of the trained MLNN, millimeter wave (mmWave) radar point cloud data representing fall attributes from monitoring the subject via mmWave radar sensing device, the input layers associated with the fall attributes, receiving, at a second set of input layers, personal attributes of the subject associated with ones of the second set of input layers, the first and second sets of input layers interconnected with an output layer of the trained MLNN via intermediate layers, the trained MLNN produced by establishing a correlation between an injury condition of prior subjects and mmWave point cloud data and personal attributes associated with the prior subjects, and generating, at the output layer, the fall injury condition attributable to the subject.

An apparatus includes a memory and processor. The memory stores OCR and NLP algorithms. The processor receives an image of a physical document page and executes the OCR algorithm to convert the image into text. The processor identifies errors in the text, which are associated with noise in the image. The processor generates a feature vector that includes features obtained by executing the NLP algorithm on the text, and features associated with the identified errors in the text. The processor uses the feature vector to assign the image to a document category. Documents assigned to the document category share one or more characteristics, and the feature vector is associated with a probability greater than a threshold that the physical document associated with the image includes those characteristics. The processor then stores the image in a database as a page of an electronic document belonging to the assigned document category.

A method for a plausibility check of the output of an artificial neural network (ANN) utilized as a classifier. The method includes: a plurality of images for which the ANN has ascertained an association with one or multiple classes of a predefined classification, and the association that is ascertained in each case by the ANN, are provided; for each image at least one feature parameter is determined which characterizes the type, the degree of specificity, and/or the position of at least one feature contained in the image; for each combination of an image and an association, a spatially resolved relevance assessment of the image is ascertained by applying a relevance assessment function; a setpoint relevance assessment is ascertained for each combination, using the feature parameter; a quality criterion for the relevance assessment function is ascertained based on the agreement between the relevance assessments and the setpoint relevance assessments.