Accelerating OpenVX through Halide and MLIR

In recent years, as many social media and AI-enabled applications have become increasingly ubiquitous, camera-centric applications have emerged as the most popular category of apps on mobile phones. A programmer can develop a camera application in an hour or less without any knowledge related to thi...

Full description

Saved in:
Bibliographic Details
Published in:Journal of signal processing systems Vol. 95; no. 5; pp. 571 - 584
Main Authors: Wu, Song-Lin, Wang, Xiang-Yu, Peng, Ming-Yi, Liao, Shih-Wei
Format: Journal Article
Language:English
Published: New York Springer US 01.05.2023
Springer Nature B.V
Subjects:
Application
Binary codes
Cameras
Cellular telephones
Circuits and Systems
Computer Imaging
Computer vision
Decoupling
Deep learning
Dialects
Electrical Engineering
Embedded systems
Engineering
Frame analysis
Image Processing and Computer Vision
Language
Linear algebra
Mobile phones
Pattern Recognition
Pattern Recognition and Graphics
Personal computers
Property
Scheduling
Signal,Image and Speech Processing
Social media
Subject specialists
Vision
Halide
OpenVX
MLIR
Image Processing
ISSN:1939-8018, 1939-8115
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In recent years, as many social media and AI-enabled applications have become increasingly ubiquitous, camera-centric applications have emerged as the most popular category of apps on mobile phones. A programmer can develop a camera application in an hour or less without any knowledge related to this domain by using different API provided by frameworks. This allows for the rapid promotion of this technology. OpenVX is a computer vision framework with vital considerations for performance and portability. This paper proposes a new framework that effectively accelerates OpenVX with Halide and MLIR. Our framework possesses Halide’s properties of decoupling algorithms and also has schedules such as an auto-scheduler. It also has MLIR’s multi-level dialects that structure the operations and the data accesses. To generate more efficient programs, we propose a bridge that can transform the programs written in OpenVX into Halide and then translate from Halide to MLIR. In the process, Our framework attains both Halide’s scheduling and MLIR’s dialect to generate more efficient binary code for execution speed.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1939-8018
1939-8115
DOI:10.1007/s11265-022-01826-8