본문 바로가기
KERD
논문KCI 등재

임베디드 시스템에서의 객체 분류를 위한 TVM기반의 성능 최적화 연구

TVM-based Performance Optimization for Image Classification in Embedded Systems

이대우|허청환|예민해|신익희
대한임베디드공학회 (Institute of Embedded Engineering of Korea)|2023.06|대한임베디드공학회논문지 (IEMEK Journal of Embedded Systems and Applications)|제18권 제3호|pp. 101-108|DOI 10.14372/IEMEK.2023.18.3.101

0

조회

0

다운로드

0

인용

0

참고문헌

키워드

Artificial IntelligenceDeep LearningEmbedded SystemQuantizationCode optimizationTVM

초록

Optimizing the performance of deep neural networks on embedded systems is a challenging task that requires efficient compilers and runtime systems. We propose a TVM-based approach that consists of three steps: quantization, auto-scheduling, and ahead-of-time compilation. Our approach reduces the computational complexity of models without significant loss of accuracy, and generates optimized code for various hardware platforms. We evaluate our approach on three representative CNNs using ImageNet Dataset on the NVIDIA Jetson AGX Xavier board and show that it outperforms baseline methods in terms of processing speed.

Abstract

Optimizing the performance of deep neural networks on embedded systems is a challenging task that requires efficient compilers and runtime systems. We propose a TVM-based approach that consists of three steps: quantization, auto-scheduling, and ahead-of-time compilation. Our approach reduces the computational complexity of models without significant loss of accuracy, and generates optimized code for various hardware platforms. We evaluate our approach on three representative CNNs using ImageNet Dataset on the NVIDIA Jetson AGX Xavier board and show that it outperforms baseline methods in terms of processing speed.

참고문헌 (35)

  1. R. Padilla, "A Survey on Performance Metrics for Object-detection Algorithms", 2020 International Conference on Systems, Signals and Image Processing (IWSSIP), pp.237~242 (2020)
  2. A. Kumar, "Face Detection Techniques: A Review", Artificial Intelligence Review, 52, pp.927~948 (2019)
  3. Y. Kang, "Natural Language Processing (NLP) in Management Research:A Literature Review", Journal of Management Analytics, 7(2), pp.139~172 (2020)
  4. J ,Chen, "Deep Learning with Edge Computing:A Review", Proceedings of the IEEE, 107(8), pp.1655~1674 (2019)
  5. T. Chen, "TVM: An Automated End-to-end Optimizing Compiler for Deep Learning", ~ (2018)
  6. H. Wu, "Integer Quantization for Deep Learning Inference:Principles and Empirical Evaluation", ~ (2020)
  7. M. A. C. Fernandes, "A Novel Training Strategy for Deep Learning Model Compression Applied to Viral Classifications", 2021 International Joint Conference on Neural Networks (IJCNN), pp.1~9 (2021)
  8. Y. Zhou, "QuantBayes: Weight Optimization for Memristive Neural Networks Via Quantization-aware Bayesian Inference", IEEE Transactions on Circuits and Systems I : Regular Papers, 68(12), pp.4851~4861 (2021)
  9. N. Shoghi, "SmaQ:Smart Quantization for DNN Training by Exploiting Value Clustering", IEEE Computer Architecture Letters, 20(2), pp.126~129 (2021)
  10. B. Liberatori, "YOLO-Based Face Mask Detection on Low-End Devices Using Pruning and Quantization", 2022 45th Jubilee International Convention on Information, Communication and Electronic Technology (MIPRO), pp.900~905 (2022)
  11. S. Zhang, "Neural Network Quantization Methods For Voice Wake Up Network", Journal of Physics : Conference Series, 1871(1), pp.012049~ (2021)
  13. Performance Analysis on TVM Optimization for AI Framework in Autonomous Vehicles, "M. H. Shin", Institute of Embedded Engineering of Korea, ~ (2021)
  14. L. Zheng, "Ansor:Generating High-performance Tensor Programs for Deep Learning", Proceedings of the 14th USENIX Conference on Operating Systems Design and Implementation, pp.863~879 (2020)
  15. A Study on the Effect of Low-level Code Optimization for DNNs via TVM Optimization Performance Analysis, "M. H. Shin", Korea Institute of Military Science and Technology, ~ (2021)
  16. M. H. Shin, "A Study on TVM for the Embedded Software in Weapon Systems", The Korea Institute of Intelligent Transport Systems, 2022(6), pp.246~251 (2022)
  17. H. A. Abdelhafez, "Snowflakes at the Edge: A Study of Variability Among NVIDIA Jetson AGX Xavier Boards", Proceedings of the 4th International Workshop on Edge Systems, Analytics and Networking, pp.1~6 (2021)
  18. K. He, "Deep Residual Learning for Image Recognition", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.770~778 (2016)
  19. M. Sandler, "Mobilenetv2: Inverted Residuals and Linear Bottlenecks", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.4510~4520 (2018)
  20. M. Tan, "Efficientnetv2: Smaller Models and Faster Training", International Conference on Machine Learning, pp.10096~10106 (2021)
  21. J. Deng, "Imagenet: A Large-scale Hierarchical Image Database", 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp.248~255 (2009)
  22. Y. Wei-Wei, "Real-Time Drivers' Violation Behaviors Detection Based on Improved YOLOv3-tiny Algorithm-Based on Model Pruning and Half-Precision Acceleration", Computer Systems & Applications, 29(04), pp.41~47 (2020)
  23. D. Lin, "Fixed Point Quantization of Deep Convolutional Networks", International Conference on Machine Learning, pp.2849~2858 (2016)
  24. P. Nayak, "Bit Efficient Quantization for Deep Neural Networks", 2019 Fifth Workshop on Energy Efficient Machine Learning and Cognitive Computing-NeurIPS Edition (EMC2-NIPS), pp.52~56 (2019)
  25. S. Yang, "Evolutionary Clustering Based Vector Quantization and SPIHT Coding for Image Compression", Pattern Recognition Letters, 31(13), pp.1773~1780 (2010)
  26. R. C. O. Rocha, "Vectorization-aware Loop Unrolling with Seed Forwarding", Proceedings of the 29th International Conference on Compiler Construction, pp.1~13 (2020)
  27. K. Hammond, "Profiling Scheduling Strategies on the GRIP Parallel Reducer", Submitted to Journal of Parallel and Distributed Computing (1991)
  28. S. D. Hammond, "Sandia ATDM Performance Execution Tools & Analysis", ~ (2018)
  29. M. Abadi, "Tensorflow: Large-scale Machine Learning on Heterogeneous Distributed Systems", ~ (2016)
  30. A. Paszke, "Pytorch: An Imperative Style, High-performance Deep Learning Library", Advances in Neural Information Processing Systems, 32 (2019)
  31. J. Bai, "Onnx: Open Neural Network Exchange", pp.54~ (2019), GitHub Repository
  33. G. Bradski, "The OpenCV Library", Dr. Dobb's Journal : Software Tools for the Professional Programmer, 25(11), pp.120~123 (2000)
  34. "S32V Vision and Sensor Fusion Evaluation Board"[URL]
  35. D. H. Son, "Development of High Reliable Real-Time Operating System (RTWORKS)Based on Partitioning and Application of Weapon System", Communications of the Korean Institute of Information Scientists and Engineers, 34(10), pp.53~59 (2016)

관련 논문AI추천

임베디드 시스템에서의 객체 분류를 위한 인공 신경망 경량화 연구신희중, 오현동한국로봇학회 · 로봇학회 논문지2022 · 제17권 제2호임베디드 엣지 플랫폼에서의 경량 비전 트랜스포머 성능 평가김태현, 이민하, 이성재대한임베디드공학회 · 대한임베디드공학회논문지2023 · 제18권 제3호FPGA기반 뉴럴네트워크 가속기에서 2차 타일링 기반 행렬 곱셈 최적화권용인, 권진세, 김태호, 김형신, 박제만, 유미선, 이제민대한임베디드공학회 · 대한임베디드공학회논문지2022 · 제17권 제6호심층 신경망을 위한 클러스터링 기반 모델 압축 기법채병철, 허선영한국정보처리학회 · 정보처리학회 논문지2024 · 제13권 제11호Edge TPU에서의 실시간성 보장을 위한 실시간 DNN 프레임워크오상은, 한창헌한국차세대컴퓨팅학회 · 한국차세대컴퓨팅학회 논문지2022 · 제18권 제5호Cycle-accurate NPU 시뮬레이터 및 데이터 접근 방식에 따른 NPU 성능평가권구윤, 박상우, 서태원대한임베디드공학회 · 대한임베디드공학회논문지2022 · 제17권 제4호

이 간행물의 다른 논문 (5건)

임베디드 엣지 플랫폼에서의 경량 비전 트랜스포머 성능 평가김태현, 이성재, 이민하pp. 89-100식품 및 축산물 금속검출기를 위한 네트워크 기반 자동화 시스템 설계조항석, 이동익pp. 109-116특이값 분해를 이용한 치수측정 기반 디지털 트윈 알고리즘 경량화박대진, 이승민pp. 117-124드론 영상을 이용한 딥러닝 기반 회전 교차로 교통 분석 시스템황윤호, 최지원, 이종택, 권희정, 이장훈pp. 125-132하천 범람 및 차량 침수 가능성 예측을 통한 딥러닝 기반 차수막 자동화 시스템함승재, 정성우, 강민수, 유준혁pp. 133-139
원문 보기