{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T15:01:22Z","timestamp":1777388482009,"version":"3.51.4"},"reference-count":63,"publisher":"Association for Computing Machinery (ACM)","issue":"5","funder":[{"DOI":"10.13039\/501100003451","name":"University of the Basque Country","doi-asserted-by":"crossref","award":["GIU21\/007"],"award-info":[{"award-number":["GIU21\/007"]}],"id":[{"id":"10.13039\/501100003451","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100003086","name":"Basque Government","doi-asserted-by":"crossref","award":["PRE_2024_2_0154"],"award-info":[{"award-number":["PRE_2024_2_0154"]}],"id":[{"id":"10.13039\/501100003086","id-type":"DOI","asserted-by":"crossref"}]},{"name":"Union Europea-NextGenerationEU through the C\u00e1tedras Chip program","award":["SOC4SENSING TSI-069100-2023-0004"],"award-info":[{"award-number":["SOC4SENSING TSI-069100-2023-0004"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2025,9,30]]},"abstract":"<jats:p>The use of hyperspectral imaging (HSI) for autonomous navigation is a promising field of research that aims at improving the accuracy and robustness of detection, tracking, and scene understanding systems based on vision sensors. The combination of advanced computer algorithms, such as deep neural networks (DNNs), and small-size snapshot HSI cameras allows to strengthen the reliability of those vision systems. Using HSI, some intrinsic limitations of greyscale and RGB imaging in depicting physical properties of targets related to the spectral reflectance of materials (metamerism) are overcome. Despite the promising results of many published HSI-based computer vision developments, the strict requirements of safety-critical applications such as autonomous driving systems (ADS) regarding latency, resource consumption, and security are prompting the migration of machine learning (ML)-based solutions to edge platforms. This involves a thorough software\/hardware co-design scheme to distribute and optimize the tasks efficiently among the limited resources of computing platforms. With respect to inference, the over-parameterized nature of DNNs poses significant computational challenges for real-time on-the-edge deployment. In addition, the intensive data preprocessing required by HSI, which is frequently overlooked, must be carefully managed in terms of memory arrangement and inter-task communication to enable an efficient integrated pipeline design on a system on chip (SoC). This work presents a set of optimization techniques for the practical co-design of a DNN-based HSI segmentation processor deployed on a field programmable gate array (FPGA)-based SoC targeted at ADS, including key optimizations such as functional software\/hardware task distribution, hardware-aware preprocessing, ML model compression, and a complete pipelined deployment. Applied compression techniques significantly reduce the complexity of the designed DNN to 24.34% of the original operations and to 1.02% of the original number of parameters, achieving a 2.86\u00d7 speed-up in the inference task without noticeable degradation of the segmentation accuracy.<\/jats:p>","DOI":"10.1145\/3748722","type":"journal-article","created":{"date-parts":[[2025,7,16]],"date-time":"2025-07-16T11:07:56Z","timestamp":1752664076000},"page":"1-27","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Optimization of DNN-based HSI Segmentation FPGA-based SoC for ADS: A Practical Approach"],"prefix":"10.1145","volume":"24","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6633-4148","authenticated-orcid":false,"given":"Jon","family":"Guti\u00e9rrez-Zaballa","sequence":"first","affiliation":[{"name":"Electronics Technology \/ Bilbao School of Engineering, University of the Basque Country (UPV\/EHU)","place":["Bilbao, Spain"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5934-4735","authenticated-orcid":false,"given":"Koldo","family":"Basterretxea","sequence":"additional","affiliation":[{"name":"Electronics Technology \/ Bilbao School of Engineering, University of the Basque Country (UPV\/EHU)","place":["Bilbao, Spain"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1064-2555","authenticated-orcid":false,"given":"Javier","family":"Echanobe","sequence":"additional","affiliation":[{"name":"Electricity and Electronics \/ Faculty of Science and Technology, University of the Basque Country (UPV\/EHU)","place":["Leioa, Spain"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2025,9,13]]},"reference":[{"key":"e_1_3_1_2_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2015.7298965"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.neucom.2022.01.005"},{"key":"e_1_3_1_4_2","doi-asserted-by":"publisher","DOI":"10.3390\/s22249790"},{"key":"e_1_3_1_5_2","doi-asserted-by":"publisher","DOI":"10.3390\/s22030757"},{"key":"e_1_3_1_6_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.tifs.2021.02.044"},{"key":"e_1_3_1_7_2","doi-asserted-by":"publisher","DOI":"10.1364\/JOSAA.23.002359"},{"key":"e_1_3_1_8_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.sysarc.2023.102878"},{"key":"e_1_3_1_9_2","doi-asserted-by":"publisher","DOI":"10.1109\/WHISPERS61460.2023.10430802"},{"key":"e_1_3_1_10_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW63382.2024.00143"},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.1117\/12.2077583"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/WHISPERS52202.2021.9483975"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1117\/12.2037607"},{"key":"e_1_3_1_14_2","article-title":"Kria K26 SOM: The Ideal Platform for Vision AI at the Edge","year":"2021","unstructured":"AMD-Xilinx. 2021. Kria K26 SOM: The Ideal Platform for Vision AI at the Edge. Retrieved April 28, 2025 from https:\/\/docs.xilinx.com\/v\/u\/en-US\/wp529-som-benchmarks. (2021).","journal-title":"R"},{"key":"e_1_3_1_15_2","article-title":"Kria K26 SOM Data Sheet (DS987)","year":"2024","unstructured":"AMD-Xilinx. 2024. Kria K26 SOM Data Sheet (DS987). Retrieved April 28, 2025 from https:\/\/docs.amd.com\/r\/en-US\/ds987-k26-som\/Overview. (2024).","journal-title":"R"},{"key":"e_1_3_1_16_2","article-title":"Arm Cortex-A53 MPCore Processor Technical Reference Manual","year":"2018","unstructured":"ARM. 2018. Arm Cortex-A53 MPCore Processor Technical Reference Manual. Retrieved November 08, 2024 from https:\/\/developer.arm.com\/documentation\/ddi0500. (2018).","journal-title":"https:\/\/developer.arm.com\/documentation\/ddi0500"},{"key":"e_1_3_1_17_2","article-title":"DPUCZDX8G for Zynq UltraScale+ MPSoCs Product Guide (PG338)","year":"2024","unstructured":"Xilinx. 2024. DPUCZDX8G for Zynq UltraScale+ MPSoCs Product Guide (PG338). Retrieved April 28, 2025 from https:\/\/docs.amd.com\/r\/en-US\/pg338-dpu","journal-title":"R"},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.1109\/SSCI52147.2023.10371793"},{"key":"e_1_3_1_19_2","doi-asserted-by":"publisher","unstructured":"Koldo Basterretxea Jon Guti\u00e9rrez-Zaballa Javier Echanobe and Mar\u00eda Victoria Mart\u00ednez. 2023. HSI-Drive. (2023). DOI:10.5281\/zenodo.15686957","DOI":"10.5281\/zenodo.15686957"},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCVW.2017.39"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1109\/IROS58592.2024.10801768"},{"key":"e_1_3_1_23_2","doi-asserted-by":"publisher","unstructured":"Shaodi You Erqi Huang Shuaizhe Liang Yongrong Zheng Yunxiang Li Fan Wang Sen Lin Qiu Shen Xun Cao Diming Zhang et\u00a0al. 2019. Hyperspectral city V1.0 dataset and benchmark. arXiv (2019). DOI:10.48550\/arXiv.1907.10270","DOI":"10.48550\/arXiv.1907.10270"},{"key":"e_1_3_1_24_2","doi-asserted-by":"publisher","unstructured":"Yuxing Huang Tianqi Ren Qiu Shen Ying Fu and Shaodi You. 2021. HSICityV2: Urban scene understanding via hyperspectral images. (Jul2021). DOI:10.5281\/zenodo.7030857","DOI":"10.5281\/zenodo.7030857"},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICME46284.2020.9102890"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1109\/IV48863.2021.9575298"},{"key":"e_1_3_1_27_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.350"},{"key":"e_1_3_1_28_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.00271"},{"key":"e_1_3_1_29_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.patrec.2008.04.005"},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.534"},{"key":"e_1_3_1_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW.2018.00141"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01385"},{"key":"e_1_3_1_33_2","unstructured":"Cityscapes Dataset. 2024. Cityscapes Dataset Benchmarks. (2024). Retrieved September 16 2024 from https:\/\/www.cityscapes-dataset.com\/benchmarks\/"},{"key":"e_1_3_1_34_2","doi-asserted-by":"crossref","unstructured":"Christoph H\u00fcmmer Manuel Schwonberg Liangwei Zhou Hu Cao Alois Knoll and Hanno Gottschalk. 2024. Strong but simple: A baseline for domain generalized dense perception by clip-based transfer learning. In Proceedings of the Asian Conference on Computer Vision. 4223\u20134244.","DOI":"10.1007\/978-981-96-0972-7_27"},{"key":"e_1_3_1_35_2","unstructured":"Michael Zhu and Suyog Gupta. 2017. To prune or not to prune: Exploring the efficacy of pruning for model compression. arXiv:1710.01878. Retrieved from https:\/\/arxiv.org\/abs\/1710.01878"},{"key":"e_1_3_1_36_2","doi-asserted-by":"publisher","DOI":"10.1007\/s11760-020-01760-x"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.01152"},{"key":"e_1_3_1_38_2","first-page":"4475","volume-title":"Proceedings of the Advances in Neural Information Processing Systems.","volume":"35","author":"Frantar Elias","year":"2022","unstructured":"Elias Frantar and Dan Alistarh. 2022. Optimal brain compression: A framework for accurate post-training quantization and pruning. In Proceedings of the Advances in Neural Information Processing Systems.S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35, Curran Associates, Inc., 4475\u20134488."},{"key":"e_1_3_1_39_2","first-page":"24101","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"35","author":"Kwon Woosuk","year":"2022","unstructured":"Woosuk Kwon, Sehoon Kim, Michael W. Mahoney, Joseph Hassoun, Kurt Keutzer, and Amir Gholami. 2022. A fast post-training pruning framework for transformers. In Proceedings of the Advances in Neural Information Processing Systems. S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35, Curran Associates, Inc., 24101\u201324116."},{"key":"e_1_3_1_40_2","volume-title":"Proceedings of the 12th International Conference on Learning Representations","author":"Zhang Yingtao","year":"2024","unstructured":"Yingtao Zhang, Haoli Bai, Haokun Lin, Jialin Zhao, Lu Hou, and Carlo Vittorio Cannistraci. 2024. Plug-and-play: An efficient post-training pruning method for large language models. In Proceedings of the 12th International Conference on Learning Representations."},{"key":"e_1_3_1_41_2","doi-asserted-by":"publisher","unstructured":"Jonathan Frankle and Michael Carbin. 2018. The lottery ticket hypothesis: Finding sparse trainable neural networks. arXiv (2018). DOI:10.48550\/arXiv.1803.03635","DOI":"10.48550\/arXiv.1803.03635"},{"key":"e_1_3_1_42_2","doi-asserted-by":"publisher","unstructured":"Zhuang Liu Mingjie Sun Tinghui Zhou Gao Huang and Trevor Darrell. 2018. Rethinking the value of network pruning. arXiv (2018). DOI:10.48550\/arXiv.1810.05270","DOI":"10.48550\/arXiv.1810.05270"},{"key":"e_1_3_1_43_2","article-title":"Deconstructing lottery tickets: Zeros, signs, and the supermask","volume":"32","author":"Zhou Hattie","year":"2019","unstructured":"Hattie Zhou, Janice Lan, Rosanne Liu, and Jason Yosinski. 2019. Deconstructing lottery tickets: Zeros, signs, and the supermask. Advances in Neural Information Processing Systems 32 (2019).","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.01191"},{"key":"e_1_3_1_45_2","first-page":"6682","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Malach Eran","year":"2020","unstructured":"Eran Malach, Gilad Yehudai, Shai Shalev-Schwartz, and Ohad Shamir. 2020. Proving the lottery ticket hypothesis: Pruning is all you need. In Proceedings of the International Conference on Machine Learning. PMLR, 6682\u20136691."},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.neucom.2025.129382"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2024.3447085"},{"key":"e_1_3_1_48_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-16788-1_2"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01528"},{"key":"e_1_3_1_50_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2022.102583"},{"key":"e_1_3_1_51_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP43922.2022.9747634"},{"key":"e_1_3_1_52_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01478"},{"key":"e_1_3_1_53_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.sysarc.2024.103242"},{"key":"e_1_3_1_54_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICECS58634.2023.10382745"},{"key":"e_1_3_1_55_2","unstructured":"ARM. 2020. Learn the Architecture: Introducing Neon. Retrieved November 08 2024 from https:\/\/developer.arm.com\/documentation\/102474\/0100"},{"key":"e_1_3_1_56_2","article-title":"Deep Learning with INT8 Optimization on Xilinx Devices","author":"Fu Yao","year":"2016","unstructured":"Yao Fu, Ephrem Wu, Ashish Sirasao, Sedny Attia, Kamran Khan, and Ralph Wittig. 2016. Deep Learning with INT8 Optimization on Xilinx Devices. Retrieved September 16, 2024 from https:\/\/docs.xilinx.com\/v\/u\/en-US\/wp486-deep-learning-int8. (2016).","journal-title":"R"},{"key":"e_1_3_1_57_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00951"},{"key":"e_1_3_1_58_2","volume-title":"Proceedings of the Advances in Neural Information Processing Systems.","volume":"2","author":"LeCun Yann","year":"1989","unstructured":"Yann LeCun, John Denker, and Sara Solla. 1989. Optimal brain damage. In Proceedings of the Advances in Neural Information Processing Systems.D. Touretzky (Ed.), Vol. 2, Morgan-Kaufmann."},{"key":"e_1_3_1_59_2","unstructured":"Hao Li Asim Kadav Igor Durdanovic Hanan Samet and Hans Peter Graf. 2017. Pruning Filters for Efficient ConvNets. arXiv:1608.08710. Retrieved from https:\/\/arxiv.org\/abs\/1608.08710"},{"key":"e_1_3_1_60_2","article-title":"Vitis AI User Guide. UG1414 (v3.5) Retrieved September 28, 2023 from","year":"2023","unstructured":"AMD-Xilinx. 2023. Vitis AI User Guide. UG1414 (v3.5) Retrieved September 28, 2023 from https:\/\/docs.xilinx.com\/r\/en-US\/ug1414-vitis-ai\/Vitis-AI-Overview. (2023).","journal-title":"https:\/\/docs.xilinx.com\/r\/en-US\/ug1414-vitis-ai\/Vitis-AI-Overview"},{"key":"e_1_3_1_61_2","doi-asserted-by":"publisher","DOI":"10.3390\/app11198802"},{"key":"e_1_3_1_62_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01234-2_49"},{"key":"e_1_3_1_63_2","doi-asserted-by":"publisher","DOI":"10.1109\/99.660313"},{"key":"e_1_3_1_64_2","doi-asserted-by":"publisher","DOI":"10.5555\/263953"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3748722","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,13]],"date-time":"2025-09-13T13:44:26Z","timestamp":1757771066000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3748722"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,13]]},"references-count":63,"journal-issue":{"issue":"5","published-print":{"date-parts":[[2025,9,30]]}},"alternative-id":["10.1145\/3748722"],"URL":"https:\/\/doi.org\/10.1145\/3748722","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,9,13]]},"assertion":[{"value":"2024-12-05","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-07-07","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-09-13","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}