Systems, apparatuses, and methods include technology that generates final frame predictions for a first plurality of frames of a video, where the first plurality of frames is associated with unlabeled data. The technology predicts an ordered list of actions for the first plurality of frames based on the final frame predictions, and temporally aligning the ordered list of actions to the final frame predictions to generate labels.

A method for generating synthetic images, each image simulating an image of an individual acquired by a vehicle interior camera, including: generating a plurality of models of individuals, each model including a three-dimensional representation of an individual's head; receiving a set of variable parameters and a probability distribution associated with each parameter; generating a set of configurations, each configuration corresponding to a combination of values or states taken by each parameter, such that the set of configurations is representative of the probability distribution of each parameter; generating, for each model of an individual, a set of images simulating images of the model of an individual acquired by a vehicle interior camera, where each image corresponds to a configuration generated for a variable parameter, and where each image further includes the three-dimensional positions of a set of points characterizing the individual's head; and storing all the images in a memory.

A method for generating synthetic images, each image simulating an image of an individual acquired by a vehicle interior camera, including: generating a plurality of models of individuals, each model including a three-dimensional representation of an individual's head; receiving a set of variable parameters and a probability distribution associated with each parameter; generating a set of configurations, each configuration corresponding to a combination of values or states taken by each parameter, such that the set of configurations is representative of the probability distribution of each parameter; generating, for each model of an individual, a set of images simulating images of the model of an individual acquired by a vehicle interior camera, where each image corresponds to a configuration generated for a variable parameter, and where each image further includes the three-dimensional positions of a set of points characterizing the individual's head; and storing all the images in a memory.

Embodiments of the present disclosure relate to a model optimization method, an electronic device, and a computer program product. This method includes: determining an initial learning rate combination for a deep learning model, wherein the initial learning rate combination includes a plurality of learning rates, each learning rate being determined for one of a plurality of layers of the deep learning model, and the plurality of learning rates including static learning rates and dynamic learning rates; and adjusting the initial learning rate combination to obtain a target learning rate combination, wherein an accuracy rate achieved when the target learning rate combination is used to train the deep learning model is higher than or equal to a first threshold accuracy rate. With the technical solution of the present disclosure, a deep learning model can be optimized by setting learning rates for each layer of the deep learning model.

The image processing device 1X includes a detection model evaluation means 31X and a display control means 33X. The detection model evaluation means 31X is configured to perform an evaluation on suitability of a detection model for detecting an attention area to be noted based on a captured image Ic in which an inspection target is photographed by a photographing unit provided in an endoscope. The display control means 33X is configured to display candidate area information according to a display mode determined based on a result of the evaluation, the candidate area information indicating one or more candidate areas that are one or more candidates of the attention area, the candidate areas being detected by one or more detection models included in detection model(s) subjected to the evaluation, the candidate area information being superimposed on the captured image Ic which is displayed by a display device 2X.

The image processing device 1X includes a detection model evaluation means 31X and a display control means 33X. The detection model evaluation means 31X is configured to perform an evaluation on suitability of a detection model for detecting an attention area to be noted based on a captured image Ic in which an inspection target is photographed by a photographing unit provided in an endoscope. The display control means 33X is configured to display candidate area information according to a display mode determined based on a result of the evaluation, the candidate area information indicating one or more candidate areas that are one or more candidates of the attention area, the candidate areas being detected by one or more detection models included in detection model(s) subjected to the evaluation, the candidate area information being superimposed on the captured image Ic which is displayed by a display device 2X.

An information processing device that includes: an image acquisition unit that acquires a catheter image obtained by an image acquisition catheter inserted into a first cavity; and a first classification data output unit configured to input the acquired catheter image to a first classification trained model that, upon receiving input of the catheter image, outputs first classification data in which a non-biological tissue region including a first inner cavity region that is inside the first cavity and a second inner cavity region that is inside a second cavity where the image acquisition catheter is not inserted and a biological tissue region are classified as different regions, and outputs the first classification data, in which the first classification trained model is generated using first training data that indicates at least the non-biological tissue region including the first inner cavity region and the second inner cavity region and the biological tissue region.

An information processing device that includes: an image acquisition unit that acquires a catheter image obtained by an image acquisition catheter inserted into a first cavity; and a first classification data output unit configured to input the acquired catheter image to a first classification trained model that, upon receiving input of the catheter image, outputs first classification data in which a non-biological tissue region including a first inner cavity region that is inside the first cavity and a second inner cavity region that is inside a second cavity where the image acquisition catheter is not inserted and a biological tissue region are classified as different regions, and outputs the first classification data, in which the first classification trained model is generated using first training data that indicates at least the non-biological tissue region including the first inner cavity region and the second inner cavity region and the biological tissue region.

Aspects of the present disclosure involve a system comprising a computer-readable storage medium storing a program and a method for performing operations comprising: receiving an image that includes a depiction of a real-world environment; processing the image to obtain data indicating presence of a real-world object in the real-world environment; receiving input that selects an AR experience comprising an AR object; determining that the real-world object detected in the real-world environment depicted in the image indicated in the obtained data corresponds to the AR object; applying a machine learning technique to the image to generate a new image that depicts the real-world environment without the real-world object; and applying the AR object to the new image to generate a modified new image that depicts the real-world environment including the AR object in place of the real-world object.

Aspects of the present disclosure involve a system comprising a computer-readable storage medium storing a program and a method for performing operations comprising: receiving an image that includes a depiction of a real-world environment; processing the image to obtain data indicating presence of a real-world object in the real-world environment; receiving input that selects an AR experience comprising an AR object; determining that the real-world object detected in the real-world environment depicted in the image indicated in the obtained data corresponds to the AR object; applying a machine learning technique to the image to generate a new image that depicts the real-world environment without the real-world object; and applying the AR object to the new image to generate a modified new image that depicts the real-world environment including the AR object in place of the real-world object.