Video upsampling using one or more neural networks
Granted: July 16, 2024
Patent Number:
12039694
Apparatuses, systems, and techniques to enhance video are disclosed. In at least one embodiment, one or more neural networks are used to create a higher resolution video using upsampled frames from a lower resolution video.
3D surface structure estimation using neural networks for autonomous systems and applications
Granted: July 16, 2024
Patent Number:
12039663
In various examples, to support training a deep neural network (DNN) to predict a dense representation of a 3D surface structure of interest, a training dataset is generated using a parametric mathematical modeling. A variety of synthetic 3D road surfaces may be generated by modeling a 3D road surface using varied parameters to simulate changes in road direction and lateral surface slope. In an example embodiment, a synthetic 3D road surface may be created by modeling a longitudinal 3D…
Light-resampling with surface similarity test
Granted: July 16, 2024
Patent Number:
12039633
Devices, systems, and techniques to incorporate lighting effects into computer-generated graphics. In at least one embodiment, a graphical frame depicting a virtual scene comprising is rendered by generating a record indicative of one or more lights in the virtual scene, and using the record to render a pixel. A second record, indicative of other lights in the virtual scene, is selected to combine with the first record, based at least in part on similarity between surfaces associated…
Stereo depth estimation using deep neural networks
Granted: July 16, 2024
Patent Number:
12039436
Various examples of the present disclosure include a stereoscopic deep neural network (DNN) that produces accurate and reliable results in real-time. Both LIDAR data (supervised training) and photometric error (unsupervised training) may be used to train the DNN in a semi-supervised manner. The stereoscopic DNN may use an exponential linear unit (ELU) activation function to increase processing speeds, as well as a machine learned argmax function that may include a plurality of…
System and method for procedurally synthesizing datasets of objects of interest for training machine-learning models
Granted: July 16, 2024
Patent Number:
12039423
The disclosure provides a vehicle, a machine vision system, and a robot. In one example, the vehicle includes one or more processing units to implement a machine-learning model and to identify one or more objects depicted in one or more images using the machine-learning model, the machine-learning model being trained using a training dataset that includes a plurality of images procedurally synthesized according to one or more training image definitions.
Processing interrupt requests for autonomous systems and applications
Granted: July 16, 2024
Patent Number:
12039362
In various examples, a timer component that generates an event when an interrupt request has not yet been cleared within at least a predetermined amount of time.
Video upsampling using one or more neural networks
Granted: July 9, 2024
Patent Number:
12033301
Apparatuses, systems, and techniques to enhance video are disclosed. In at least one embodiment, one or more neural networks are used to create a higher resolution video using upsampled frames from a lower resolution video.
Superpixel generation and use
Granted: July 9, 2024
Patent Number:
12034967
Apparatuses, systems, and techniques to interpolate one or more intermediate images from two or more images is disclosed. In at least one embodiment, a processor includes one or more circuits to interpolate one or more intermediate images from two or more images based, at least in part, on one or more inconsistent flow vectors corresponding to the two or more images.
Receiver and transmitter adaptation using stochastic gradient hill climbing with genetic mutation
Granted: July 9, 2024
Patent Number:
12034576
A receiver receives communications over a communication channel, which may distort an incoming communication signal. In order to counter this distortion, the frequency response of the receiver is manipulated by adjusting several parameters. Each parameter controls at least a portion of the frequency response of the receiver. The optimal values for the parameters are determined by modifying an initial set of values for the parameters through one or more stochastic hill climbing operations…
Determining high-interest durations of gameplay sessions from user inputs
Granted: July 9, 2024
Patent Number:
12033667
In various examples, durations of relatively high user activity within a gameplay session may be determined from user input events using a running user activity measurement. Once a duration is identified, it may be further analyzed to merge the duration with one or more other durations and/or to determine or predict whether the duration would be of sufficient interest for further action. A user interest score for an identified duration may be computed based on a set of the user input…
Asynchronous accumulator using logarithmic-based arithmetic
Granted: July 9, 2024
Patent Number:
12033060
Neural networks, in many cases, include convolution layers that are configured to perform many convolution operations that require multiplication and addition operations. Compared with performing multiplication on integer, fixed-point, or floating-point format values, performing multiplication on logarithmic format values is straightforward and energy efficient as the exponents are simply added. However, performing addition on logarithmic format values is more complex. Conventionally,…
System level hardware mechanisms for dynamic assist control
Granted: July 9, 2024
Patent Number:
12032840
Various embodiments include a computer memory system that dynamically adjusts a memory device performance feature, such as dynamic assist control, dynamic turbo mode, and/or the like, to improve the performance of memory devices in the memory system. The memory system enables or disables the memory device performance feature based on the operating voltage relative to a threshold voltage. If the operating voltage crosses the threshold voltage in one direction, then the memory device…
Guiding vehicles through vehicle maneuvers using machine learning models
Granted: July 9, 2024
Patent Number:
12032380
In various examples, a trigger signal may be received that is indicative of a vehicle maneuver to be performed by a vehicle. A recommended vehicle trajectory for the vehicle maneuver may be determined in response to the trigger signal being received. To determine the recommended vehicle trajectory, sensor data may be received that represents a field of view of at least one sensor of the vehicle. A value of a control input and the sensor data may then be applied to a machine learning…
Augmentation of global navigation satellite system based data
Granted: July 9, 2024
Patent Number:
12031824
A vehicle computing system validates location data received from a Global Navigation Satellite System receiver with other sensor data. In one embodiment, the system calculates velocities with the location data and the other sensor data. The system generates a probabilistic model for velocity with a velocity calculated with location data and variance associated with the location data. The system determines a confidence score by applying the probabilistic model to one or more of the…
Multi-level and multi-label content classification using unsupervised and ensemble machine learning techniques
Granted: July 2, 2024
Patent Number:
12026626
Apparatuses, systems, and techniques to classify content. In at least one embodiment, a mixture of neural networks each trained to generate labels for various types of input are used to automatically generate appropriate content labels given an input.
Mitigating duty cycle distortion degradation due to device aging on high-bandwidth memory interface
Granted: July 2, 2024
Patent Number:
12027198
Embodiments include a memory device with an improved circuit to mitigate degradation of memory devices due to aging. Memory device input/output pins include delay elements for adjusting the delay in each memory input/output signal path to synchronize the input/output signal paths with one another. Certain data patterns, including a long series of logic zero values or a long series of logic one values, can cause asymmetric degradation of transistors included in the delay elements. This…
Assigning obstacles to lanes using neural networks for autonomous machine applications
Granted: July 2, 2024
Patent Number:
12026955
In various examples, live perception from sensors of an ego-machine may be leveraged to detect objects and assign the objects to bounded regions (e.g., lanes or a roadway) in an environment of the ego-machine in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute outputs—such as output segmentation masks—that may correspond to a combination of object classification and lane identifiers. The output masks may be post-processed to determine…
Image generation using one or more neural networks
Granted: July 2, 2024
Patent Number:
12026845
Apparatuses, systems, and techniques are presented to generate augmented images. In at least one embodiment, one or more neural networks are used to modify one or more first objects in an image based at least in part upon a modification to be made to one or more second objects in the image.
Performing spherical denoising
Granted: July 2, 2024
Patent Number:
12026830
In order to perform denoising on a three-dimensional (3D) spherical measurement of light (such as spherical irradiance probe information or the results of a 3D gonioreflectometry capture), the 3D spherical measurement of light is converted to a two-dimensional (2D) measurement by creating multiple copies of the 3D spherical measurement of light, determining a two-dimensional sub-domain (e.g., a rectangular sub-domain) for each of the multiple copies, and stitching the plurality of…
Shadow denoising in ray-tracing applications
Granted: July 2, 2024
Patent Number:
12026822
In various examples, the actual spatial properties of a virtual environment are used to produce, for a pixel, an anisotropic filter kernel for a filter having dimensions and weights that accurately reflect the spatial characteristics of the virtual environment. Geometry of the virtual environment may be computed based at least in part on a projection of a light source onto a surface through an occluder, in order to determine a footprint that reflects a contribution of the light source to…
