The accurate recognition system of citrus flowers using YOLOv4-Tiny lightweight neural network and FPGA embedded platform

Shilei Lyu; Yawen Zhao; Ruiyao Li; Qiao Chen; Zhen Li

doi:10.1117/12.2611299

14 October 2021 The accurate recognition system of citrus flowers using YOLOv4-Tiny lightweight neural network and FPGA embedded platform

Shilei Lyu, Yawen Zhao, Ruiyao Li, Qiao Chen, Zhen Li

Author Affiliations +

Proceedings Volume 11930, International Conference on Mechanical Engineering, Measurement Control, and Instrumentation; 119302E (2021) https://doi.org/10.1117/12.2611299
Event: International Conference on Mechanical Engineering, Measurement Control, and Instrumentation (MEMCI 2021), 2021, Guangzhou, China

Abstract

Citrus flower recognition is the key technology of flower thinning, pollination and yield prediction in citrus orchard production and management. Current object detection algorithms are mostly limited into server computing environment, which is difficult to meet the application requirements of accurate recognition of citrus flowers in a natural environment. A citrus flower recognition model based on YOLOv4-Tiny lightweight neural network was proposed by using software and hardware co-design pattern in this paper. Through the frame transformation quantification and compilation of the recognition model, the dynamic link library was generated and transplanted to FPGA embedded platform. So, the citrus flower accurate recognition system was designed and implemented. The comprehensive recognition accuracy of citrus flower recognition model deployed on embedded platform for flowers and buds was not less than 89.30%; The size of the quantization recognition model was 5.64 MB; The accuracy loss of the model was 1.77%; The frame rate was not lower than 16 FPS. It can meet the requirements of real-time accurate identification of citrus flowers.

Citation Download Citation

Shilei Lyu, Yawen Zhao, Ruiyao Li, Qiao Chen, and Zhen Li "The accurate recognition system of citrus flowers using YOLOv4-Tiny lightweight neural network and FPGA embedded platform", Proc. SPIE 11930, International Conference on Mechanical Engineering, Measurement Control, and Instrumentation, 119302E (14 October 2021); https://doi.org/10.1117/12.2611299

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available