A method of road sign classification utilizing a knowledge graph, including detecting and selecting a representation of a sign across a plurality of frames, outputting a prompt initiating a request for a classification associated with the representation of the sign, classifying one or more images including the sign, querying the knowledge graph to obtain a plurality of road sign classes with at least one same attribute as the sign, and classifying the sign across the plurality of frames in response to a confidence level exceeding a threshold.

A system provides intelligent gallery management for biometrics. A first gallery is obtained that includes biometric and/or other information on a population of people. An application is identified. A subset of the population of people is identified based on the application. A second gallery is derived from the first gallery by pulling the information for the subset of the population of people without pulling the information for the population of people not in the subset. Biometric identification (such as facial recognition) for the application may then be performed using the second gallery rather than the first gallery. In this way, the system is improved as less time is required for biometric identification, fewer device resources are used, and so on.

Described is a system for learning object labels for control of an autonomous platform. Pseudo-task optimization is performed to identify an optimal pseudo-task for each source model of one or more source models. An initial target network is trained using the optimal pseudo-task. Source image components are extracted from source models, and an attribute dictionary of attributes is generated from the source image components. Using zero-shot attribution distillation, the unlabeled target data is aligned with the source models similar to the unlabeled target data. The unlabeled target data are mapped onto attributes in the attribute dictionary. A new target network is generated from the mapping, and the new target network is used to assign an object label to an object in the unlabeled target data. The autonomous platform is controlled based on the object label.

Aspects and embodiments relate to a method of providing a representation of a feature identified by a deep neural network as being relevant to an outcome, a computer program product and apparatus configured to perform that method. The method comprises: providing the deep neural network with a training library comprising: a plurality of samples associated with the outcome; using the deep neural network to recognise a feature in the plurality of samples associated with the outcome; creating a feature recognition library from an input library by identifying one or more elements in each of a plurality of samples in the input library which trigger recognition FIGURE is of the feature by the deep neural network; using the feature recognition library to synthesise a plurality of one or more elements of a sample which have characteristics which trigger recognition of the feature by the deep neural network; and using the synthesised plurality of one or more elements to provide a representation of the feature identified by the deep neural network in the plurality of samples associated with the outcome. Accordingly, rather than visualising a single instance of one or more elements in a sample which trigger a feature associated with an outcome, it is possible to visualise a range of samples including elements which would trigger a feature associated with an outcome, thus enabling a more comprehensive view of operation of a deep neural network in relation to a particular feature.

Aspects and embodiments relate to a method of providing a representation of a feature identified by a deep neural network as being relevant to an outcome, a computer program product and apparatus configured to perform that method. The method comprises: providing the deep neural network with a training library comprising: a plurality of samples associated with the outcome; using the deep neural network to recognise a feature in the plurality of samples associated with the outcome; creating a feature recognition library from an input library by identifying one or more elements in each of a plurality of samples in the input library which trigger recognition FIGURE is of the feature by the deep neural network; using the feature recognition library to synthesise a plurality of one or more elements of a sample which have characteristics which trigger recognition of the feature by the deep neural network; and using the synthesised plurality of one or more elements to provide a representation of the feature identified by the deep neural network in the plurality of samples associated with the outcome. Accordingly, rather than visualising a single instance of one or more elements in a sample which trigger a feature associated with an outcome, it is possible to visualise a range of samples including elements which would trigger a feature associated with an outcome, thus enabling a more comprehensive view of operation of a deep neural network in relation to a particular feature.

Systems and methods described relate to the synthesis of content using generative models. In at least one embodiment, a score-based generative model can use a stochastic differential equation with critically-damped Langevin diffusion to learn to synthesize content. During a forward diffusion process, noise can be introduced into a set of auxiliary (e.g., “velocity”) values for an input image to learn a score function. This score function can be used with the stochastic differential equation during a reverse diffusion denoising process to remove noise from the image to generate a reconstructed version of the input image. A score matching objective for the critically-damped Langevin diffusion process can require only the conditional distribution learned from the velocity data. A stochastic differential equation-based integrator can then allow for efficient sampling from these critically-damped Langevin diffusion-based models.

In a learning that uses a machine learning model for an image, a learning method, a learning model, a classifier, and a generator in which human vision is taken into consideration are provided. The learning method learns a machine learning model that inputs or outputs image data with data for learning that includes training data subjected to a process of leaving out a component that is difficult to visually judge to reduce an information amount or generated data at a predetermined ratio.

In a learning that uses a machine learning model for an image, a learning method, a learning model, a classifier, and a generator in which human vision is taken into consideration are provided. The learning method learns a machine learning model that inputs or outputs image data with data for learning that includes training data subjected to a process of leaving out a component that is difficult to visually judge to reduce an information amount or generated data at a predetermined ratio.

The present disclosure relates to a method, device, and system for emotion intervention, as well as a computer-readable storage medium and a healing room. The emotion intervention method includes: identifying a user's emotion state according to the user's first biometric information; and recommending at least one emotion intervention corresponding to the emotion state.

A computer, including a processor and a memory, the memory including instructions to be executed by the processor to determine a second convolutional neural network (CNN) training dataset by determining an underrepresented object configuration and an underrepresented noise factor corresponding to an object in a first CNN training dataset, generate one or more simulated images including the object corresponding to the underrepresented object configuration in the first CNN training dataset by inputting ground truth data corresponding to the object into a photorealistic rendering engine and generate one or more synthetic images including the object corresponding to the underrepresented noise factor in the first CNN training dataset by processing the simulated images with a generative adversarial network (GAN) to determine a second CNN training dataset. The instructions can include further instructions to train a CNN to using the first and the second CNN training datasets.