HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC
This paper proposes HW/SW Co-design of an automatic classification system of Khalas, Khunaizi, Fardh, Qash, Naghal, and Maan dates fruit varieties in Oman. The system implements pre-processing, segmentation of the colored input images, color and shape-size features extraction followed by ANN-tansig...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
FRUCT
2020-04-01
|
Series: | Proceedings of the XXth Conference of Open Innovations Association FRUCT |
Subjects: | |
Online Access: | https://www.fruct.org/publications/fruct26/files/Amm.pdf |
id |
doaj-0709942f978c42a09b097de2e8d5fa81 |
---|---|
record_format |
Article |
spelling |
doaj-0709942f978c42a09b097de2e8d5fa812020-11-25T03:42:09ZengFRUCTProceedings of the XXth Conference of Open Innovations Association FRUCT2305-72542343-07372020-04-01261101510.23919/FRUCT48808.2020.9087548HW/SW Co-Design for Dates Classification on Xilinx Zynq SoCAhmed Chiheb Ammari0Lazhar Khriji1Medhat Awadalla2Sultan Qaboos University, OmanSultan Qaboos University, OmanSultan Qaboos University, OmanThis paper proposes HW/SW Co-design of an automatic classification system of Khalas, Khunaizi, Fardh, Qash, Naghal, and Maan dates fruit varieties in Oman. The system implements pre-processing, segmentation of the colored input images, color and shape-size features extraction followed by ANN-tansig classification. The performance of the proposed system is experimented and 97.26% highest classification accuracy are achieved. The proposed system is prototyped using a selected Zynq 7020 SoC platform featuring, on the same chip, a dual-core ARM Cortex A9 processing System (PS) interconnected with FPGA logic (PL) though high-throughput communication channels. The original classification algorithm is profiled and then a HW/SW Co-design is developed achieving 10.9 fps real time classification performance. This performance is acceptable and represents almost 14 times speedup acceleration comparatively to the original program implementation.https://www.fruct.org/publications/fruct26/files/Amm.pdfartificial neural networkcolor and shape-size featureszynq sochw/sw co-design |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Ahmed Chiheb Ammari Lazhar Khriji Medhat Awadalla |
spellingShingle |
Ahmed Chiheb Ammari Lazhar Khriji Medhat Awadalla HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC Proceedings of the XXth Conference of Open Innovations Association FRUCT artificial neural network color and shape-size features zynq soc hw/sw co-design |
author_facet |
Ahmed Chiheb Ammari Lazhar Khriji Medhat Awadalla |
author_sort |
Ahmed Chiheb Ammari |
title |
HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC |
title_short |
HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC |
title_full |
HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC |
title_fullStr |
HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC |
title_full_unstemmed |
HW/SW Co-Design for Dates Classification on Xilinx Zynq SoC |
title_sort |
hw/sw co-design for dates classification on xilinx zynq soc |
publisher |
FRUCT |
series |
Proceedings of the XXth Conference of Open Innovations Association FRUCT |
issn |
2305-7254 2343-0737 |
publishDate |
2020-04-01 |
description |
This paper proposes HW/SW Co-design of an automatic classification system of Khalas, Khunaizi, Fardh, Qash, Naghal, and Maan dates fruit varieties in Oman. The system implements pre-processing, segmentation of the colored input images, color and shape-size features extraction followed by ANN-tansig classification. The performance of the proposed system is experimented and 97.26% highest classification accuracy are achieved. The proposed system is prototyped using a selected Zynq 7020 SoC platform featuring, on the same chip, a dual-core ARM Cortex A9 processing System (PS) interconnected with FPGA logic (PL) though high-throughput communication channels. The original classification algorithm is profiled and then a HW/SW Co-design is developed achieving 10.9 fps real time classification performance. This performance is acceptable and represents almost 14 times speedup acceleration comparatively to the original program implementation. |
topic |
artificial neural network color and shape-size features zynq soc hw/sw co-design |
url |
https://www.fruct.org/publications/fruct26/files/Amm.pdf |
work_keys_str_mv |
AT ahmedchihebammari hwswcodesignfordatesclassificationonxilinxzynqsoc AT lazharkhriji hwswcodesignfordatesclassificationonxilinxzynqsoc AT medhatawadalla hwswcodesignfordatesclassificationonxilinxzynqsoc |
_version_ |
1724526921388130304 |