Apple Inc.

Hsiao-Chen Chang

Sheng Lin

Muge Wang

Weichun Ku

Ponan Kuo

Ji Liang Song

Yixiang Tao

Embodiments relate to a neural processor circuit that includes a first number of neural engine circuits, a second number of channels and a data processor circuit. The first number of neural engine circuits are pipelined into the second number of chains smaller than the first number. Each of the chains is configured to generate output data of a first size. Each of the channels is coupled to each of the chains and configured to transmit the output data from each of the neural engine circuits in the chains sequentially. The data processor circuit is coupled to the channels to receive the output data. The data processor circuit aggregates the output data of each of the chains into aggregated data of a second size larger than the first size and writes the aggregated data of the second size into a buffer memory of the data processor circuit.

Chained neural engine write-back architecture

Embodiments of the present disclosure relate to generating motion vectors. An image signal processor includes a statistics circuit and a vector correlation analysis circuit. The statistics circuit determines image statistics such as vectors representing sums of pixel values of rows or columns of blocks of an image. Additionally, the statistics circuit may mix or aggregate sums of multiple color components. The vector correlation analysis performs cross-correlation between vectors of a current image and reference vectors of a prior image to determine cross-correlation scores. The vector correlation analysis generates a motion vector by identifying shifts in horizontal and vertical directions corresponding to peak values of cross-correlation scores.

Image statistics for motion detection

Growing community of inventors

Los Altos, CA, United States of America

Hsiao-Chen Chang