Resource-constrained FPGA implementation of YOLOv2

Abstract

Progress is being made to deploy convolutional neural networks (CNNs) into the Internet of Things (IoT) edge devices for handling image analysis tasks locally. These tasks require low-latency and low-power computation on low-resource IoT edge devices. However, CNN-based algorithms, e.g. YOLOv2, typically contain millions of parameters. With the increase in the CNN’s depth, filters are increased by a power of two. A large number of filters and operations could lead to frequent off-chip memory access that affects the operation speed and power consumption of the device. Therefore, it is a challenge to map a deep CNN into a low-resource edge IoT platform. To address this challenge, we present a resource-constrained Field-Programmable Gate Array implementation of YOLOv2 with optimized data transfer and computing efficiency. Firstly, a scalable cross-layer dataflow strategy is proposed which allows on-chip data transfer between different types of layers, and offers flexible off-chip data transfer when the intermediate results are unaffordable on-chip. Next, a filter-level data-reuse dataflow strategy together with a filter-level parallel multiply-accumulate operation computing processing elements array is developed. Finally, multi-level sliding buffers are developed to optimize the convolutional computing loop and reuse the input feature maps and weights. Experiment results show that our implementation has achieved 4.8 W of low-power consumption for executing YOLOv2, an 8-bit deep CNN containing 50.6 MB weights, using low-resource of 8.3 Mbits on-chip memory. The throughput and power efficiency are 100.33 GOP/s and 20.90 GOP/s/W, respectively.

Review of Energy-Efficient Embedded System Acceleration of Convolution Neural Networks for Organic Weeding Robots

“…According to the research of [44], they presented an FPGA-based implementation of YOLOv2 for object detection tasks on resource-constrained computer vision-based IoT edge devices and remote control vehicles equipped with cameras. Their approach focuses on resolving issues related to data reloading and off-chip memory access, employing an efficient dataflow strategy and multi-level buffers that maximize on-chip data transfer and minimize external memory access.…”

Review of State-of-the-Art FPGA Applications in IoT Networks

“…To have a broader impact on IoT applications, future works should implement the aforementioned works but with more focus on power-saving features. This is especially necessary for IoT nodes that do not have access to a stable power source, and instead rely on battery or solar power [ 68 , 69 , 70 ].…”

Implementing Convolutional Neural Networks on FPGA: A Survey and Research

“…ZHANG, Zhichao, MAHMUD, MA Parvez, and KOUZANI, Abbas Z. [18] Present object Detection on Low-Resource IoT Devices using FPGA-based YOLOv2 Implementation. Improves data transfer with efficient dataflow and multi-level buffers, leading to low memory utilization and power consumption of 4.8 W. Limitations include conflict between YOLOv2 size and low-resource requirement and support for only 8-bit precision.…”