{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T16:13:33Z","timestamp":1775837613445,"version":"3.50.1"},"reference-count":29,"publisher":"Wiley","license":[{"start":{"date-parts":[[2024,4,2]],"date-time":"2024-04-02T00:00:00Z","timestamp":1712016000000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Technological Innovation and Application Development of Chongqing","award":["CSTB2022TIAD-KPX0057"],"award-info":[{"award-number":["CSTB2022TIAD-KPX0057"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["International Journal of Distributed Sensor Networks"],"published-print":{"date-parts":[[2024,4,2]]},"abstract":"<jats:p>The low-bit quantization can effectively reduce the deep neural network storage as well as the computation costs. Existing quantization methods have yielded unsatisfactory results when being applied to lightweight networks. Additionally, following network quantization, the differences in data types between the operators can cause issues when deploying networks on Field Programmable Gate Arrays (FPGAs). Moreover, some operators cannot be accelerated heterogeneously on FPGAs, resulting in frequent switching between the Advanced RISC Machine (ARM) and FPGA environments for computation tasks. To address these problems, this paper proposes a custom network quantization approach. Firstly, an improved PArameterized Clipping Activation (PACT) method is employed during the quantization aware training to restrict the value range of neural network parameters and reduce the loss of precision arising from quantization. Secondly, the Consecutive Execution Of Convolution Operators (CEOCO) strategy is utilized to mitigate the resource consumption caused by the frequent environment switching. The proposed approach is validated on Xilinx Zynq Ultrascale+MPSoC 3EG and Virtex UltraScale+XCVU13P platforms. The MobileNetv1, MobileNetv3, PPLCNet, and PPLCNetv2 networks were utilized as testbeds for the validation. Moreover, experimental results are on the miniImageNet, CIFAR-10, and OxFord 102 Flowers public datasets. In comparison to the original model, the proposed optimization methods result in an average decrease of 1.2% in accuracy. Compared to conventional quantization method, the accuracy remains almost unchanged, while the frames per second (FPS) on FPGAs improves by an average of 2.1 times.<\/jats:p>","DOI":"10.1155\/2024\/8018810","type":"journal-article","created":{"date-parts":[[2024,4,3]],"date-time":"2024-04-03T08:07:25Z","timestamp":1712131645000},"page":"1-11","source":"Crossref","is-referenced-by-count":3,"title":["Custom Network Quantization Method for Lightweight CNN Acceleration on FPGAs"],"prefix":"10.1155","volume":"2024","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-7862-910X","authenticated-orcid":true,"given":"Lingjie","family":"Yi","sequence":"first","affiliation":[{"name":"School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2986-6356","authenticated-orcid":true,"given":"Xianzhong","family":"Xie","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China"}]},{"given":"Yi","family":"Wan","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-8564-8884","authenticated-orcid":true,"given":"Bo","family":"Jiang","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-9393-3569","authenticated-orcid":true,"given":"Junfan","family":"Chen","sequence":"additional","affiliation":[{"name":"Chongqing Haiyun Jiexun Technology, Chongqing, China"}]}],"member":"311","reference":[{"key":"1","doi-asserted-by":"publisher","DOI":"10.1109\/TMM.2020.2997126"},{"key":"2","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2019.105210"},{"key":"3","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2022.3155123"},{"key":"4","doi-asserted-by":"publisher","DOI":"10.1007\/s00530-022-00973-z"},{"key":"5","doi-asserted-by":"publisher","DOI":"10.13374\/j.issn2095-9389.2019.03.27.002"},{"key":"6","doi-asserted-by":"publisher","DOI":"10.1007\/s10462-020-09816-7"},{"key":"7","doi-asserted-by":"publisher","DOI":"10.1007\/s10489-020-01894-y"},{"key":"8","doi-asserted-by":"publisher","DOI":"10.23919\/FPL.2017.8056820"},{"key":"9","article-title":"PACT: parameterized clipping activation for quantized neural networks","author":"J. Choi","year":"2018"},{"key":"10","doi-asserted-by":"crossref","first-page":"126","DOI":"10.1007\/978-3-030-68238-5_10","article-title":"Subtensor quantization for Mobilenets","volume-title":"In Computer Vision\u2013ECCV 2020 Workshops: Glasgow, UK, August 23\u201328, 2020, Proceedings, Part V 16","author":"T. Dinh","year":"2020"},{"key":"11","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-030-58536-5_5","article-title":"Post-training piecewise linear quantization for deep neural networks","volume-title":"Computer Vision\u2013ECCV 2020: 16th European Conference, Glasgow, UK, August 23\u201328, 2020, Proceedings, Part II 16, pages 69\u201386","author":"J. Fang","year":"2020"},{"key":"12","first-page":"2704","article-title":"Quantization and training of neural networks for efficient integer-arithmetic-only inference","author":"B. Jacob"},{"key":"13","first-page":"90","article-title":"Performance improvements in quantization aware training and appreciation of low precision computation in deep learning","volume-title":"Advances in Signal Processing and Intelligent Recognition Systems: 6th International Symposium, SIRS 2020, Chennai, India, October 14\u201317, 2020, Revised Selected Papers 6","author":"K. Uday","year":"2021"},{"key":"14","doi-asserted-by":"publisher","DOI":"10.1109\/ISCTech60480.2023.00024"},{"key":"15","doi-asserted-by":"publisher","DOI":"10.1177\/1550147719868669"},{"key":"16","doi-asserted-by":"publisher","DOI":"10.1007\/s00521-018-3761-1"},{"key":"17","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.3000009"},{"key":"18","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2022.3179016"},{"key":"19","doi-asserted-by":"publisher","DOI":"10.3390\/electronics10040518"},{"key":"20","article-title":"Deep compression: compressing deep neural networks with pruning, trained quantization and Huffman coding","author":"S. Han","year":"2015"},{"key":"21","first-page":"1638","article-title":"SqueezeNext: hardware-aware neural network design","author":"A. Gholami"},{"issue":"2","key":"22","doi-asserted-by":"crossref","first-page":"227","DOI":"10.3102\/1076998619872761","article-title":"Deep learning with tensorflow: a review","volume":"45","author":"B. Pang","year":"2020","journal-title":"Journal of Educational and Behavioral Statistics"},{"key":"23","article-title":"Pytorch: an imperative style, high-performance deep learning library","volume":"32","author":"A. Paszke","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"24","first-page":"578","article-title":"{TVM}: An automated {end-to-end} optimizing compiler for deep learning","volume-title":"In 13th USENIX Symposium on Operating Systems Design and Implementation (OSDI 18)","author":"T. Chen","year":"2018"},{"key":"25","doi-asserted-by":"publisher","DOI":"10.11871\/jfdc.issn.2096.742X.2019.01.011"},{"key":"26","article-title":"Matching networks for one shot learning","volume":"29","author":"O. Vinyals","year":"2016","journal-title":"Advances in Neural Information Processing Systems"},{"key":"27","doi-asserted-by":"publisher","DOI":"10.1109\/CSCI.2016.0225"},{"key":"28","doi-asserted-by":"publisher","DOI":"10.1109\/ICVGIP.2008.47"},{"key":"29","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-009-0275-4"}],"container-title":["International Journal of Distributed Sensor Networks"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/downloads.hindawi.com\/journals\/dsn\/2024\/8018810.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"http:\/\/downloads.hindawi.com\/journals\/dsn\/2024\/8018810.xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"http:\/\/downloads.hindawi.com\/journals\/dsn\/2024\/8018810.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,4,3]],"date-time":"2024-04-03T08:08:43Z","timestamp":1712131723000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.hindawi.com\/journals\/dsn\/2024\/8018810\/"}},"subtitle":[],"editor":[{"given":"Kavita","family":"Pandey","sequence":"additional","affiliation":[]}],"short-title":[],"issued":{"date-parts":[[2024,4,2]]},"references-count":29,"alternative-id":["8018810","8018810"],"URL":"https:\/\/doi.org\/10.1155\/2024\/8018810","relation":{},"ISSN":["1550-1477","1550-1329"],"issn-type":[{"value":"1550-1477","type":"electronic"},{"value":"1550-1329","type":"print"}],"subject":[],"published":{"date-parts":[[2024,4,2]]}}}