Conference proceeding
REQ-YOLO: A Resource-Aware, Efficient Quantization Framework for Object Detection on FPGAs
PROCEEDINGS OF THE 2019 ACM/SIGDA INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE GATE ARRAYS (FPGA'19), pp 33-42
01 Jan 2019
Featured in Collection : UN Sustainable Development Goals @ Drexel
Abstract
Deep neural networks (DNNs), as the basis of object detection, will play a key role in the development of future autonomous systems with full autonomy. The autonomous systems have special requirements of real-time, energy-efficient implementations of DNNs on a power-constrained system. Two research thrusts are dedicated to performance and energy efficiency enhancement of the inference phase of DNNs. The first one is model compression techniques while the second is efficient hardware implementation. Recent works on extremely-low-bit CNNs such as the binary neural network (BNN) and XNOR-Net replace the traditional floating point operations with binary bit operations which significantly reduces the memory bandwidth and storage requirement. However, it suffers from non-negligible accuracy loss and underutilized digital signal processing (DSP) blocks of FPGAs.
To overcome these limitations, this paper proposes REQ-YOLO, a resource aware, systematic weight quantization framework for object detection, considering both algorithm and hardware resource aspects in object detection. We adopt the block-circulant matrix method and propose a heterogeneous weight quantization using Alternative Direction Method of Multipliers (ADMM), an effective optimization technique for general, non-convex optimization problems. To achieve real-time, highly-efficient implementations on FPGA, we present the detailed hardware implementation of block circulant matrices on CONV layers and develop an efficient processing element (PE) structure supporting the heterogeneous weight quantization, CONV dataflow and pipelining techniques, design optimization, and a template-based automatic synthesis framework to optimally exploit hardware resource. Experimental results show that our proposed REQ-YOLO framework can significantly compress the YOLO model while introducing very small accuracy degradation. The related codes are here: https://github.com/Anonymous788/heterogeneous_ADMM_YOLO.
Metrics
Details
- Title
- REQ-YOLO: A Resource-Aware, Efficient Quantization Framework for Object Detection on FPGAs
- Creators
- Caiwen Ding - Northeastern UniversityShuo Wang - Peking UniversityNing Liu - Northeastern UniversityKaidi Xu - Northeastern UniversityYanzhi Wang - Northeastern UniversityYun Liang - Peng Cheng LaboratoryACM
- Publication Details
- PROCEEDINGS OF THE 2019 ACM/SIGDA INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE GATE ARRAYS (FPGA'19), pp 33-42
- Publisher
- Assoc Computing Machinery
- Number of pages
- 10
- Grant note
- Z181100008918015 / Municipal Science and Technology Program 1704662; 1739748 / National Science Foundation; National Science Foundation (NSF) L172004 / Beijing Natural Science Foundation 1704662 / Direct For Computer & Info Scie & Enginr; Division Of Computer and Network Systems; National Science Foundation (NSF); NSF - Directorate for Computer & Information Science & Engineering (CISE)
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Computer Science
- Web of Science ID
- WOS:000522383700005
- Scopus ID
- 2-s2.0-85064404653
- Other Identifier
- 991021871490104721
UN Sustainable Development Goals (SDGs)
This publication has contributed to the advancement of the following goals:
InCites Highlights
Data related to this publication, from InCites Benchmarking & Analytics tool:
- Collaboration types
- Domestic collaboration
- International collaboration
- Web of Science research areas
- Computer Science, Theory & Methods