基于硬件仿真系统的边缘计算人工智能视觉芯片设计验证

doi:10.11959/j.issn.2096-3750.2022.00250

Abstract

Abstract:

The rise of visual deep learning algorithms based on convolutional neural network (CNN) has promoted the rapid development of the artificial intelligence (AI) vision chip design research.The step of chip verification is a bottleneck in the development of AI vision chips.A software and hardware verification method for AI vision chip design based on hardware simulation system was introduced.Taking AI vision chip design for edge computing as an example, the chip was run on the hardware simulation system (ZeBu) and the simulation verification work of typical deep learning network MobileNet was completed.The results show that the network model implemented on the hardware chip architecture keeps accuracy while the detection time of a single frame is only 18.51 ms under a 200 MHz clock frequency.The spread of the hardware simulation is 7 times faster than than of the software simulation.

Key words: AI vision chip, deep learning, MobileNet, ZeBu

CLC Number:

TN47

Xuanzhe XU, Ke NING, Xuemin ZHENG, Mingxin ZHAO, Mengmeng XU, Nanjian WU, Liyuan LIU. Verification of an artificial intelligence vision chip design for edge computing based on hardware simulation system[J]. Chinese Journal on Internet of Things, 2022, 6(1): 20-28.

Figures/Tables 11

References 25

[1]	KRIZHEVSK Y A , SUTSKEVER I , HINTON G E . ImageNet classification with deep convolutional neural networks[J]. Communication softhe ACM, 2017,60(6): 84-90.
[2]	SZEGED Y C , LIU W , JIA Y Q ,et al. Going deeper with convolutions[C]// Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2015: 1-9.
[3]	DONG X D , XU Y P , XU Z J ,et al. A static hand gesture recognition model based on the improved centroid watershed algorithm and a dual-channel CNN[C]// Proceedings of 2018 24th International Conference on Automation and Computing (ICAC). Piscataway:IEEEPress, 2018: 1-6.
[4]	REN S Q , HE K M , GIRSHICK R ,et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,39(6): 1137-1149.
[5]	SHIN H C , ROTH H R , GAO M C ,et al. Deep convolutional neural networks for computer-aided detection:CNN architectures,dataset characteristics and transfer learning[J]. IEEE Transactions on Medical Imaging, 2016,35(5): 1285-1298.
[6]	LIN T Y , ROYCHOWDHURY A , MAJI S . Bilinear CNN models for fine-grained visual recognition[C]// Proceedings of 2015 IEEE International Conference on Computer Vision. Piscataway:IEEE Press, 2015: 1449-1457.
[7]	WE IH , ZHU M , WANG B ,et al. Two-level progressive attention convolutional network for fine-grained image recognition[J]. IEEE Access, 2020(8): 104985-104995.
[8]	ZHANG Y , YANG S Y , LI H B ,et al. Shadow tracking of moving target based on CNN for video SAR system[C]// Proceedings of IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium. Piscataway:IEEE Press, 2018: 4399-4402.
[9]	WANG M , ABDELFATTAH S , MOUSTAFA N ,et al. Deep Gaussian mixture-hidden Markov model for classification of EEG signals[J]. IEEE Transactions on Emerging Topics in Computational Intelligence, 2018,2(4): 278-287.
[10]	CHEN L , ZHANG H W , XIAO J ,et al. SCA-CNN:spatial and channel-wise attention in convolutional networks for image captioning[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2017: 6298-6306.
[11]	ZHANG X Y , ZHOU X Y , LIN M X ,et al. ShuffleNet:an extremely efficient convolutional neural network for mobile devices[C]// Proceedings of 2018 IEEE/CVFConference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2018: 6848-6856.
[12]	SUDHA S , JAYANTHI K B , RAJASEKARAN C ,et al. Segmentation of RoI in medical images using CNN-A comparative study[C]// Proceedings of TENCON 2019 - 2019 IEEE Region 10 Conference. Piscataway:IEEE Press, 2019: 767-771.
[13]	SHARMA A K , FOROOSH H . Slim-CNN:alight-weight CNN for face attribute prediction[C]// Proceedings of 2020 15th IEEE International Conference on Automatic Face and Gesture Recognition. Piscataway:IEEE Press, 2020: 329-335.
[14]	HOWARD A G , ZHU M L , CHEN B ,et al. MobileNets:efficient convolutional neural networks for mobile vision applications[EB]. 2017.
[15]	SHI C , YANG J , HAN Y ,et al. A 1000 fps vision chip based on a dynamically reconfigurable hybrid architecture comprising a PE array processor and self-organizing map neural network[J]. IEEE Journal of Solid-State Circuits, 2014,49(9): 2067-2082.
[16]	LIHL , ZHANGZ X , YANG J ,et al. A novel vision chip architecture for image recognition based on convolutional neural network[C]// Proceedings of 2015 IEEE 11th International Conference on ASIC. Piscataway:IEEE Press, 2015: 1-4.
[17]	VINAY B K , HARIHARM , KILLEDAR A . The FPGA based emulation of complex SoC for ADAS market on ZeBu-Server[C]// Proceedings of 2014 International Conference on Advances in Electronics Computers and Communications. Piscataway:IEEE Press, 2014: 1-4.
[18]	ENDC. Synopsys推出业界最快的仿真系统[EB]. 2014.
	ENDC. Synopsys Iaunches the industry’s fastest simulation system[EB]. 2014.
[19]	WANG S H , JIANG Y Y , HOUXX ,et al. Cerebral micro-bleed detection based on the convolution neural network with rank based average pooling[J]. IEEE Access, 2017(5): 16576-16583.
[20]	BASHA S H S , DUBEY S R , PULABAIGARI V ,et al. Impact of fully connected layers on performance of convolutional neural networks for image classification[J]. Neurocomputing, 2020,378: 112-119.
[21]	ZEILER M D , FERGUS R . Visualizing and understanding convolutional networks[C]// Computer Vision – ECCV 2014, 2014: 818-833.
[22]	SIMONYANK ZISSERMAN A . Very deep convolutional networks for large-scale image recognition[EB]. 2014.
[23]	SZEGEDY C , VANHOUCKEV , IOFFE S ,et al. Rethinking the inception architecture for computer vision[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2016: 2818-2826.
[24]	HE K M , ZHANG X Y , REN S Q ,et al. Deep residual learning for image recognition[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2016: 770-778.
[25]	ZOPH B , VASUDEVAN V , SHLENS J ,et al. Learning transferable architectures for scalable image recognition[C]// Proceedings of 2018 IEEE/CVFConference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2018: 8697-8710.

Metrics

Recommended 0

No Suggested Reading articles found!

层数	类型	输出特征图大小（尺寸）	卷积核大小（尺寸）
1	4-D卷积层	0.438 MB（112px×128 px×32 px）	3.375 KB（3×3×3×32）
2	深度卷积层	0.438 MB（112 px×128 px×32 px）	1.125 KB（3×3×32）
3	点式卷积层	0.875 MB（112 px×128 px×64 px）	8 KB（1×1×32×64）
4	深度卷积层	0.219MB（56 px×64 px×64 px）	2.25 KB（3×3×64）
5	点式卷积层	0.438 MB（56 px×64 px×128 px）	32 KB（1×1×64×128）
6	深度卷积层	0.438 MB（56 px×64 px×128 px）	4.5 KB（3×3×128）
7	点式卷积层	0.438 MB（56 px×64 px×128 px）	64 KB（1×1×128×128）
8	深度卷积层	0.109 MB（28 px×32 px×128 px）	4.5 kB（3×3×128）
9	点式卷积层	0.219 MB（28 px×32 px×256 px）	128 KB（1×1×128×256）
10	深度卷积层	0.219 MB（28 px×32 px×256 px）	9 KB（3×3×256）
11	点式卷积层	0.219 MB（28 px×32 px×256 px）	256 KB（1×1×256×256）
12	深度卷积层	0.055 MB（14 px×16 px×256 px）	9 KB（3×3×256）
13	点式卷积层	0.109 MB（14 px×16 px×512 px）	512 KB（1×1×256×512）
14～23	深度卷积层点式卷积层	0.109 MB（14 px×16 px×512 px）0.109 MB（14 px×16 px×512 px）	18 KB（3×3×512）1 024 KB（1×1×512×512）
24	深度卷积层	0.027 MB（7 px×8 px×512 px）	18 KB（3×3×512）
25	点式卷积层	0.055 MB（7 px×8 px×1 024 px）	2 048 KB（1×1×512×1024）
26	深度卷积层	0.055 MB（7 px×8 px×1 024 px）	36 KB（3×3×1024）
27	点式卷积层	0.055 MB（7 px×8 px×1 024 px）	4 096 KB（1×1×1024×1024）
28	平均池化层	0.001 MB（1 px×1 px×1 024 px）	-
29	全连接层	0.001 MB（1 px×1 px×1 024 px）	4 096 KB（1×1×1024×1024）

序号	网络名称	参数量/MB
1	AlexNet^[1]	60
2	ZFNet^[21]	62.3
3	VGG16^[22]	138.3
4	VGG19^[22]	143.6
5	Inception V3^[23]	23.8
6	ResNet50^[24]	25.5
7	MobileNet V1^[14]	16
8	NASNetMobile^[25]	5
9	轻量化MobileNet V1（本文）	2.9

算法类型	单幅片上运行时钟周期数	单幅EDA运行时间/min	单幅ZeBu运行时间/min
图像分类	3 702 028	114.20	15.73
人脸检测	12 241 338	377.65	51.85
目标追踪	4 996 608	152.85	21.23
TOF成像	4 560 400	140.65	19.33

Verification of an artificial intelligence vision chip design for edge computing based on hardware simulation system

RichHTML

PDF下载

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 25

Related Articles 9

Metrics

Recommended 0

[1]	Zhifei ZHANG, Feng LIU, Yiyang GE, Shuo LI, Yu ZHANG, Ke XIONG. An intrusion detection method based on depthwise separable convolution and attention mechanism [J]. Chinese Journal on Internet of Things, 2023, 7(1): 49-59.
[2]	Rui JIANG, Liuting SUN, Xiaoming WANG, Dapeng LI, Youyun XU. Research on EEG signal classification of motor imagery based on AE and Transformer [J]. Chinese Journal on Internet of Things, 2023, 7(1): 118-128.
[3]	Biao ZHANG, Ximing WANG, Yifan XU, Wen LI, Hao HAN, Songyi LIU, Xueqiang CHEN. Multi-domain collaborative anti-jamming based on multi-agent deep reinforcement learning [J]. Chinese Journal on Internet of Things, 2022, 6(4): 104-116.
[4]	Dan LUO, Ruzhi XU, Zhitao GUAN. Differential privacy budget optimization based on deep learning in IoT [J]. Chinese Journal on Internet of Things, 2022, 6(2): 65-76.
[5]	Guoquan LI, Yonghai XU, Jinzhao LIN, Zhengwen HUANG. Res-DNN based signal detection algorithm for end-to-end MIMO systems [J]. Chinese Journal on Internet of Things, 2022, 6(1): 65-72.
[6]	Ling TAN, Shanshan RONG, Jingming XIA, Sarker SAJIB, Wenjie MA. Real-time diagnosis of multi-category skin diseases based on IR-VGG [J]. Chinese Journal on Internet of Things, 2021, 5(3): 115-125.
[7]	Chunmin LIN, Liekang ZENG, Xu CHEN. Research on power efficient autonomous UAV navigation algorithm: an edge intelligence driven approach [J]. Chinese Journal on Internet of Things, 2021, 5(2): 87-96.
[8]	Muhan CHEN,Jiajia GUO,Xiao LI,Shi JIN. An overview of the CSI feedback based on deep learning for massive MIMO systems [J]. Chinese Journal on Internet of Things, 2020, 4(1): 33-44.
[9]	Yong LIAO,Haimei YAO,Yuanxiao HUA. Channel state information acquisition algorithm based on deep learning for IoT [J]. Chinese Journal on Internet of Things, 2019, 3(1): 8-13.