Ittiam Systems (p) Ltd.

Xilinx, Inc.

Arun Kumar Patil

Preethi Konda

Tharun Battula

Sriram Sethuraman

Sumit Nagpal

Vishal K Jain

Abid Karumannil

Scheduling work of a machine learning application includes instantiating kernel objects by a computer processor in response to input of kernel definitions. Each kernel object is of a kernel type indicating a compute circuit. The computer processor generates a graph in a memory. Each node represents a task and specifies an assignment of the task to one or more of the kernel objects, and each edge represents a data dependency. Task queues are created in the memory and assigned to queue tasks represented by the nodes. Kernel objects are assigned to the task queues, and the tasks are enqueued by threads executing the kernel objects, based on assignments of the kernel objects to the task queues and assignments of the tasks to the kernel objects. Tasks are dequeued by the threads, and the compute circuits are activated to initiate processing of the dequeued tasks.

Scheduling processing of machine learning tasks on heterogeneous compute circuits

A system and method for tone mapping on a high dynamic range (HDR) image of a scene are disclosed. In one embodiment, tone mapping on the HDR image is performed using a set of bit-depths, that corresponds to a plurality of images of the scene, obtained at each unit of an irradiance range of the scene.

System and method for tone mapping on high dynamic range images

A method for generating a high dynamic range (HDR) image from images of a scene obtained at one or more exposure values is disclosed. In this embodiment, one of the obtained images is selected as a reference image. Further, mapped images are obtained by mapping pixel intensity values to corresponding irradiance values in each image. Furthermore, a pixel intensity value dependent weighting factor is determined. Moreover, a set of images is identified from the mapped images for pixels in the reference image. Also, a set of corresponding mapped pixels is established for the mapped pixels in reference image in the set of images. Further, a similarity measure is computed for the mapped pixels of reference image and corresponding mapped pixels in the set of images. Furthermore, each mapped pixel of the reference image is combined with a subset of its established corresponding mapped pixels in the set of images.

System and method for generating high dynamic range images

Growing community of inventors

Bengaluru, India

Arun Kumar Patil