Texas Instruments Corporation

Kumar Arrakutti Desappan

Pramod Kumar Swami

Manu Mathew

Mihir Narendra Mody

Soyeb Noormohammed Nagori

Yashwant Dutt

Piyali Goswami

Prashanth Ramanathpur Viswanath

Praveen Eppa

Deepak Kumar Poddar

Hrushikesh Tukaram Garud

Kedar Satish Chitnis

Shyam Jagannathan

David E Smith

Rishabh Garg

Anshu Jain

Techniques for executing machine learning (ML) models including receiving an indication to run an ML model on a processing core; receiving a static memory allocation for running the ML model on the processing core; determining that a layer of the ML model uses more memory than the static memory allocated; transmitting, to a shared memory, a memory request for blocks of the shared memory; receiving an allocation of the requested blocks; running the layer of the ML model using the static memory and the range of memory addresses; and outputting results of running the layer of the ML model.

Memory sharing for machine learning processing

Techniques including receiving a first set of values for processing by a machine learning (ML) network, storing a first portion of the first set of values in an on-chip memory, processing the first portion of the first set of values in a first layer of the ML network to generate a second portion of a second set of values, overwriting the stored first portion with the generated second portion, processing the second portion in a second layer of the ML network to generate a third portion of a third set of values, storing the third portion, repeating the steps of storing the first portion, processing the first portion, overwriting the stored first portion, processing the second portion, and storing the third portion for a fourth portion of the first set of values until all portions of the first set of values are processed to generate the third set of values.

Super-tiling in neural network processing to enable analytics at lower memory speed

A method for video object detection includes detecting an object in a first video frame, and selecting a first interest point and a second interest point of the object. The first interest point is in a first region of interest located at a first corner of a box surrounding the object. The second interest point is in a second region of interest located at a second corner of the box. The second corner is diagonally opposite the first corner. A first optical flow of the first interest point and a second optical flow of the second interest point are determined. A location of the object in a second video frame is estimated by determining, in the second video frame, a location of the first interest point based on the first optical flow and a location of the second interest point based on the second optical flow.

Video object detection

This description relates to image feature extraction. In some examples, a system can include a keypoint detector and a feature list generator. The keypoint detector can be configured to upsample a keypoint score map to produce an upsampled keypoint score map. The keypoint score map can include feature scores indicative of a likelihood of at least one feature being present at keypoints in an image. The feature list generator can be configured to identify a subset of keypoints of the keypoints in the image using the feature scores of the up sampled keypoint score map, determine descriptors for the subset of keypoints based on a feature description map, and generate a keypoint descriptor map for the image based on the determined descriptors.

Systems and methods for image feature extraction

A method for dynamically quantizing feature maps of a received image. The method includes convolving an image based on a predicted maximum value, a predicted minimum value, trained kernel weights and the image data. The input data is quantized based on the predicted minimum value and predicted maximum value. The output of the convolution is computed into an accumulator and re-quantized. The re-quantized value is output to an external memory. The predicted min value and the predicted max value are computed based on the previous max values and min values with a weighted average or a pre-determined formula. Initial min value and max value are computed based on known quantization methods and utilized for initializing the predicted min value and predicted max value in the quantization process.

Dynamic quantization for deep neural network inference system and method

Several methods and systems for masking multimedia data are disclosed. In an embodiment, a method for masking includes performing a prediction for at least one multimedia data block based on a prediction mode of a plurality of prediction modes. The at least one multimedia data block is associated with a region of interest (ROI). A residual multimedia data associated with the at least one multimedia data block is generated based on the prediction. A quantization of the residual multimedia data is performed based on a quantization parameter (QP) value. The QP value is variable such that varying the QP value controls a degree of masking of the ROI.

Methods and systems for masking multimedia data

A method for convolution in a convolutional neural network (CNN) is provided that includes accessing a coefficient value of a filter corresponding to an input feature map of a convolution layer of the CNN, and performing a block multiply accumulation operation on a block of data elements of the input feature map, the block of data elements corresponding to the coefficient value, wherein, for each data element of the block of data elements, a value of the data element is multiplied by the coefficient value and a result of the multiply is added to a corresponding data element in a corresponding output block of data elements comprised in an output feature map.

Reduced complexity convolution for convolutional neural networks

Systems and methods are provided for selecting feature points within an image. A plurality of candidate feature points are identified in the image. A plurality of feature points are selected for each of the plurality of candidate feature points, a plurality of sets of representative pixels. For each set of representative pixels, a representative value is determined as one of a maximum chromaticity value and a minimum chromaticity value from the set of representative pixels. A score is determined for each candidate feature point from the representative values for the plurality of sets of representative pixels associated with the candidate feature point. The feature points are selected according to the determined scores for the plurality of candidate feature points.

Efficient feature point selection

An object detection system and a method of detecting an object in an image are disclosed. In an embodiment, a method for detecting the object includes computing one or more feature planes of one or more types for each image pixel of the image. A plurality of cells is defined in the image, where each cell includes first through nnumber of pixels, and starting locations of each cell in the image in horizontal and vertical directions are integral multiples of predefined horizontal and vertical step sizes, respectively. One or more feature plane summations of one or more types are computed for each cell. A feature vector is determined for an image portion of the image based on a set of feature plane summations, and the feature vector is compared with a corresponding object classifier to detect a presence of the corresponding object in the image portion of the image.

Systems and methods for object detection

A video system includes an encoder for generating a compressed bit stream in response to a received video signal. A transitional location is located in the bit stream that includes skip macro-block encoding, where the transitional location provides an indication where a linear relationship starts to exist between a length of the encoded bit stream and a length of a number of skipped macro-blocks. The skipped macro-blocks in the bit stream before the transitional location are encoded using a predetermined nonlinear relationship between the length of the encoded bit stream and the length of the number of skipped macro-blocks, whereas skipped macro-blocks in the bit stream are encoded using a linear relationship between the length of the encoded bit stream and the length of the number of skipped macro-blocks.

Growing community of inventors

Ekambara Kuppam, India

Kumar Arrakutti Desappan