Xperi develops world first neuromorphic in-cabin sensing powered by the efficiency and accuracy of Prophesee Metavision®
www.prophesee.ai
Talk about the minotaur's cave Nothing for Xperi, nothing for DTS, so I tried CTO Petronel Bigioi - Found a few for FOTONATION LTD:
FotoNation is a wholly owned subsidiary of Xperi.
US11046327B2 System for performing eye detection and/or tracking
[0034] A
s further illustrated in FIG. 3, the system 302 may include a face detector component 316 , a control component 318 , and an eye tracking component 320 . The face detector component 316 may be configured to analyze the first image data 310 in order to determine a location of a face of a user. For example, the face detector component 316 may analyze the first image data 310 using one or more algorithms associated with face detection. The one or more algorithms may include, but are not limited to, neural network algorithm(s), Principal Component Analysis algorithm(s), Independent Component Analysis algorithms(s), Linear Discriminant Analysis algorithm(s), Evolutionary Pursuit algorithm(s), Elastic Bunch Graph Matching algorithm(s), and/or any other type of algorithm(s) that the face detector component 316 may utilize to perform face detection on the first image data 310 .
[0041] The eye tracking component
320 may be configured to analyze the second image data
312 in order to determine eye position and/or a gaze direction of the user. For example, the eye tracking component
320 may analyze the second image data
312 using one or more algorithms associated with eye tracking. The one or more algorithms may include, but are not limited to, neural network algorithm(s) and/or any other types of algorithm(s) associated with eye tracking.
Missed it by that much:
[0050] As
described herein, a machine-learned model which may include, but is not limited to a neural network (e.g., You Only Look Once (YOLO) neural network, VGG, DenseNet, PointNet, convolutional neural network (CNN), stacked auto-encoders, deep Boltzmann machine (DBM), deep belief networks (DBN),), regression algorithm (e.g., ordinary least squares regression (OLSR), linear regression, logistic regression, stepwise regression, multivariate adaptive regression splines (MARS), locally estimated scatterplot smoothing (LOESS)), Bayesian algorithms (e.g., naïve Bayes, Gaussian naïve Bayes, multinomial naïve Bayes, average one-dependence estimators (AODE), Bayesian belief network (BNN), Bayesian networks), clustering algorithms (e.g., k-means, k-medians, expectation maximization (EM), hierarchical clustering), association rule learning algorithms (e.g., perceptron, back-propagation, Hopfield network, Radial Basis Function Network (RBFN)), supervised learning, unsupervised learning, semi-supervised learning, etc. Additional or alternative examples of neural network architectures may include neural networks such as ResNet50, ResNet101, VGG, DenseNet, PointNet, and the like. Although discussed in the context of neural networks, any type of machine-learning may be used consistent with this disclosure. For example, machine-learning algorithms may include, but are not limited to, regression algorithms, instance-based algorithms, Bayesian algorithms, association rule learning algorithms, deep learning algorithms, etc.
... no suggestion of a digital SNN SoC, or even an analog one.
However, they did make a CNN SoC:
WO2017129325A1 A CONVOLUTIONAL NEURAL NETWORK
Von Neumann rools!
A
convolutional neural network (CNN) for an image processing system comprises an image cache responsive to a request to read a block of NxM pixels extending from a specified location within an input map to provide a block of NxM pixels at an output port. A convolution engine reads blocks of pixels from the output port, combines blocks of pixels with a corresponding set of weights to provide a product, and subjects the product to an activation function to provide an output pixel value. The image cache comprises a plurality of interleaved memories capable of simultaneously providing the NxM pixels at the output port in a single clock cycle. A controller provides a set of weights to the convolution engine before processing an input map, causes the convolution engine to scan across the input map by incrementing a specified location for successive blocks of pixels and generates an output map within the image cache by writing output pixel values to successive locations within the image cache.
