Abstract: The present disclosure relates to a custom framework for fine-grained human activity recognition. One or more input videos may be accessed, where the one or more input videos comprise one or more frames depicting one or more actors and one or more objects. A plurality of object-pose interaction graphs may be generated for individual frames from the one or more input videos based at least in part on one or more objects of interest from the one or more objects and on one or more joint keypoints of the one or more actors. A first graph neural network may be trained based at least in part on the plurality of object-pose interaction graphs to identify spatial information for the one or more actors, the one or more objects of interest, and one or more interactions between the one or more actors and the one or more objects of interest.

Abstract: Techniques are described herein for generating block extender model. An example method includes a system accessing a base model trained for identifying a base class. The system can access an extender comprising block extenders, the extender class distinct from the base class. The system can connect the extender with the base model to generate an augmented model. The system can input training data to the augmented model, the training data is provided to the base model and the extender, the training data comprising data points associated with the extender class. The system can train the extender model to identify the extender class based at least in part on the training data and the signal received from the base machine learning model. The system can generate a trained extender based at least in part on the training, the extender trained to identify an object associated with the extender class.

Abstract: Continual learning techniques are described for extending the capabilities of a base model, which is trained to predict a set of existing or base classes, to generate a target model that is capable of making predictions for both the existing or base classes and additionally for making predictions for new or custom classes. The techniques described herein enable the target model to be trained such that the model can make predictions involving both base classes and custom classes with high levels of accuracy.

Abstract: Techniques and systems for classifying non-detected faults (NDFs) in a formal verification test-bench are described. A sequential equivalence checking formulation can be constructed based on an integrated circuit (IC) design and a set of NDFs, wherein the set of NDFs do not falsify a first set of properties of the IC design, wherein said constructing the sequential equivalence checking formulation comprises creating a second set of properties based on the set of NDFs, wherein each property in the second set of properties corresponds to an NDF in the set of NDFs. A formal sequential equivalence checking tool can be used to prove the second set of properties in the sequential equivalence checking formulation. Next, for each property in the second set of properties that is disproven by the formal sequential equivalence checking tool, some embodiments can classify a corresponding NDF in the set of NDFs as an observable NDF.

Abstract: In one aspect, a fault injection environment and a formal property verification environment are combined in a single integrated flow that allows the user to go back and forth between the two tasks. A system that unifies formal property verification and fault injection includes user interfaces that support the unified use model. In one approach, the FPV tool is the master and its user interface is the primary interface for the user to set up, run and debug faults as well as checkers. This interface allows the user to interactively select the FPV properties and/or the faults to be used for fault analysis. The user interface may provide a view of the faults, for example by listing faults or summarizing faults by class, type, etc.

Abstract: An apparatus and method are provided for handling 360 degree image content captured by a 360 degree camera. A method includes obtaining, by an electronic device, a sequence of 360 degree projection format image frames associated with the 360 degree image content and inertial measurement unit (IMU) rotation data associated with the 360 degree image content, computing, by the electronic device, a relative rotation of the 360 degree image content, and applying, by the electronic device, the relative rotation to at least one of the 360 degree projection format image frames to stabilize the 360 degree image content.

Abstract: A method of selecting capture configuration based on scene analysis and an image capturing device are provided. The method includes analyzing by a processor, a scene currently being captured by an image capturing device having a plurality of imaging sensors, identifying a current image capturing mode of the image capturing device, setting at least one capture parameter for the plurality of the imaging sensors, upon determining that the at least one capture parameter of the current image capturing mode has to be changed, and determining a timing sequence for triggering the plurality of imaging sensors to capture a plurality of image frames based on the set at least one capture parameter.

Abstract: The fault analysis problem is modelled by automatically creating additional properties (fault properties) and constraints based on a plurality of injected faults and existing user assertions. These fault properties and constraints are sent to formal verification in a single run to qualify all of the faults together, rather than sequentially checking each fault in a separate formal verification run.

Abstract: An apparatus and method are provided for handling 360 degree image content captured by a 360 degree camera. A method includes obtaining, by an electronic device, a sequence of 360 degree projection format image frames associated with the 360 degree image content and inertial measurement unit (IMU) rotation data associated with the 360 degree image content, computing, by the electronic device, a relative rotation of the 360 degree image content, and applying, by the electronic device, the relative rotation to at least one of the 360 degree projection format image frames to stabilize the 360 degree image content.

Abstract: In one aspect, a fault injection environment and a formal property verification environment are combined in a single integrated flow that allows the user to go back and forth between the two tasks. A system that unifies formal property verification and fault injection includes user interfaces that support the unified use model. In one approach, the FPV tool is the master and its user interface is the primary interface for the user to set up, run and debug faults as well as checkers. This interface allows the user to interactively select the FPV properties and/or the faults to be used for fault analysis. The user interface may provide a view of the faults, for example by listing faults or summarizing faults by class, type, etc.

Abstract: The fault analysis problem is modelled by automatically creating additional properties (fault properties) and constraints based on the injected faults and the existing user assertions. These fault properties and constraints are sent to formal verification in a single run to qualify all of the faults together, rather than sequentially checking each fault in a separate formal verification run.

Abstract: A method of selecting capture configuration based on scene analysis and an image capturing device are provided. The method includes analyzing by a processor, a scene currently being captured by an image capturing device having a plurality of imaging sensors, identifying a current image capturing mode of the image capturing device, setting at least one capture parameter for the plurality of the imaging sensors, upon determining that the at least one capture parameter of the current image capturing mode has to be changed, and determining a timing sequence for triggering the plurality of imaging sensors to capture a plurality of image frames based on the set at least one capture parameter.