Abstract: Techniques are disclosed for performing residual image compression techniques used in conjunction with image and/or video predictors. The techniques utilize a compression scheme that implements a noise model to estimate noise values of pixels in an originally acquired image. These noise value estimates are then used to perform residual image compression more efficiently by performing a non-uniform reduction in resolution of the residual image. The resolution reduction includes dropping least significant bits (LSBs) used to encode each pixel on a pixel-by-pixel basis based upon the noise value estimates of the originally acquired image.

Abstract: Techniques are disclosed for improving the manner in which multiple HDR images are generated having different exposure times. The technicism allow for the use of a single imaging sensor to generate HDR images with a reduced latency required to do so. The result is that both long and high exposure HDR images may be generated in a much faster time frame than that required for traditional HDR sensors, which may be a time offset of the shorter exposure time of the two HDR images. This allows for the images to capture more similar scenes given the proximity in time in which both are generated, allowing for more accurate vehicle-based functions to be implemented that rely upon such HDR images, such as object classification.

Abstract: A method for channel specific neural network processing includes receiving current layer multi-channel output descriptors by next layer neurons. The current layer multi-channel output descriptors are provided by neurons of the current layer. The current layer and the next layer belong to a neural network. The next layer neurons process the current layer multi-channel output descriptors to provide next layer multi-channel output descriptors. The processing includes multiplying the current layer multi-channel output descriptors by channel compensated weights of the next layer neurons to provide next layer products that compensate for estimated differences between scale factors associated with different channels of the current layer multi-channel output descriptors. The next layer products are quantized by applying next layer output channel specific quantization.

Abstract: Techniques are disclosed for the implementation of machine learning model training utilities to generate models for advanced driving assistance system (ADAS), driving assistance, and/or automated vehicle (AV) systems. The techniques described herein may be implemented in conjunction with the utilization of open source and cloud-based machine learning training utilities to generate machine learning trained models. One example of such an open source solution includes TensorFlow, which is a free and open-source software library for dataflow and differentiable programming across a range of tasks. TensorFlow may be used in conjunction with many different types of machine learning utilities.