Abstract
In this work a new approach for the large-scale face-feature matching is introduced. Based on this approach, the LSH was used to hash the database to a smaller set of face-feature vectors have the most similarities with the searched vector. This hashed database is then further processed by linear search on FPGA. Experimental implementations show that the searching results of our method have higher accuracy of 99,6% and less computational time of 43ms in comparison to other methods such as pure-LSH, or LSH combined with CPU. This approach is very promising for the problem of large-scale data matching on edge devices having very limited hardware resources.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Wang, M., Deng, W.: Deep Face Recognition: a Survey. arXiv:1804.06655 (2018)
Locality sensitive hashing. https://www.mit.edu/~andoni/LSH/. Accessed 20 Nov 2020
Andoni, A., Indyk, P., Laarhoven, T., Razenshteyn, I., Schmidt, L.: Practical and Optimal LSH for Angular Distance. arXiv:1509.02897 (2015)
FALCONN github repository. https://github.com/FALCONN-LIB/FALCONN. Accessed 24 Nov 2020
Melikoglu, O., Ergin, O., Salami, B., Pavon, J., Unsal, O., Cristal, A.: A Novel FPGA-Based High Throughput Accelerator for Binary Search Trees. arXiv:1912.01556 (2019)
Chen, H., Madaminov, S., Ferdman, M., Milder, P.: FPGA-accelerated samplesort for large data sets. In: 2020 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA 2020), pp. 222–232. Association for Computing Machinery, New York (2020)
Jun, S., Chung, C., Arvind: Large-scale high-dimensional nearest neighbor search using flash memory with in-store processing. In: 2015 International Conference on ReConFigurable Computing and FPGAs (ReConFig), pp. 1–8. IEEE, Mexico (2015)
Zhang, J., Li, J., Khoram, S.: Efficient large-scale approximate nearest neighbor search on OpenCL FPGA. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4924–4932. IEEE, Mexico (2018)
Cosine Similarity Using Xilinx Alveo. https://developer.xilinx.com/en/articles/cosine-similarity-using-xilinx-alveo.html. Accessed 20 Nov 2020
Large-scale CelebFaces Attributes (CelebA) Dataset. https://mmlab.ie.cuhk.edu.hk/projects/CelebA.html. Accessed 29 Nov 2020
Zynq UltraScale+ MPSoC ZCU104 Evaluation Kit. https://www.xilinx.com/products/boards-and-kits/zcu104.html. Accessed 29 Nov 2020
Acknowledgement
This work was funded by CMC Institute of Science and Technology, CMC Corporation, Hanoi, Vietnam.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Dong, HN., Ha, NX., Tuan, DM. (2021). A New Approach for Large-Scale Face-Feature Matching Based on LSH and FPGA for Edge Processing. In: Tran, DT., Jeon, G., Nguyen, T.D.L., Lu, J., Xuan, TD. (eds) Intelligent Systems and Networks . ICISN 2021. Lecture Notes in Networks and Systems, vol 243. Springer, Singapore. https://doi.org/10.1007/978-981-16-2094-2_42
Download citation
DOI: https://doi.org/10.1007/978-981-16-2094-2_42
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-16-2093-5
Online ISBN: 978-981-16-2094-2
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)