LMM4-IC4K: A Large Multimodal Model Powered Integrated Circuit Footprint Geometry Understanding

Yida Wang, Taiting Lu, Runze Liu, Lanqing Yang, Zhe Chen, Yuehai Wang, Yixin Liu, Kaiyuan Lin, Xiaomeng Chen, Dian Ding, Yijie Li, Yifan Yang, Yi-Chao Chen, Yincheng Jin, Mahanth Gowda
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:128742-128761, 2026.

Abstract

Printed-Circuit-board (PCB) footprint geometry labeling of integrated circuits (IC) is essential in defining the physical interface between components and the PCB layout, requiring precise visual perception. However, the unstructured nature of footprint drawings and abstract diagram annotations prevents direct IC footprint parsing and automated package geometry labeling methods from developing. Existing Large Multimodal Models (LMMs) struggle with inaccurate geometric perception, limiting their effectiveness in this task. To address these challenges, we propose LMM4-IC4K, a novel framework that treats IC mechanical drawings as images and leverages LMMs for structured geometric interpretation. To support such a framework, we introduce ICGeo8K, a multi-modal dataset with 8,608 labeled samples, including 4138 real-world IC footprint samples and 4470 synthetically generated samples. We further present a two-stage training framework to fine-tune LMMs for IC footprint labeling. Extensive experiments demonstrate that our model outperforms state-of-the-art LMMs on the proposed benchmark. The accurate translation of footprint diagrams enabled by LMM4-IC4K contributes to advancing automation and standardization within the PCB industry.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-wang26fg, title = {{LMM}4-{IC}4{K}: A Large Multimodal Model Powered Integrated Circuit Footprint Geometry Understanding}, author = {Wang, Yida and Lu, Taiting and Liu, Runze and Yang, Lanqing and Chen, Zhe and Wang, Yuehai and Liu, Yixin and Lin, Kaiyuan and Chen, Xiaomeng and Ding, Dian and Li, Yijie and Yang, Yifan and Chen, Yi-Chao and Jin, Yincheng and Gowda, Mahanth}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {128742--128761}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/wang26fg/wang26fg.pdf}, url = {https://proceedings.mlr.press/v306/wang26fg.html}, abstract = {Printed-Circuit-board (PCB) footprint geometry labeling of integrated circuits (IC) is essential in defining the physical interface between components and the PCB layout, requiring precise visual perception. However, the unstructured nature of footprint drawings and abstract diagram annotations prevents direct IC footprint parsing and automated package geometry labeling methods from developing. Existing Large Multimodal Models (LMMs) struggle with inaccurate geometric perception, limiting their effectiveness in this task. To address these challenges, we propose LMM4-IC4K, a novel framework that treats IC mechanical drawings as images and leverages LMMs for structured geometric interpretation. To support such a framework, we introduce ICGeo8K, a multi-modal dataset with 8,608 labeled samples, including 4138 real-world IC footprint samples and 4470 synthetically generated samples. We further present a two-stage training framework to fine-tune LMMs for IC footprint labeling. Extensive experiments demonstrate that our model outperforms state-of-the-art LMMs on the proposed benchmark. The accurate translation of footprint diagrams enabled by LMM4-IC4K contributes to advancing automation and standardization within the PCB industry.} }
Endnote
%0 Conference Paper %T LMM4-IC4K: A Large Multimodal Model Powered Integrated Circuit Footprint Geometry Understanding %A Yida Wang %A Taiting Lu %A Runze Liu %A Lanqing Yang %A Zhe Chen %A Yuehai Wang %A Yixin Liu %A Kaiyuan Lin %A Xiaomeng Chen %A Dian Ding %A Yijie Li %A Yifan Yang %A Yi-Chao Chen %A Yincheng Jin %A Mahanth Gowda %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-wang26fg %I PMLR %P 128742--128761 %U https://proceedings.mlr.press/v306/wang26fg.html %V 306 %X Printed-Circuit-board (PCB) footprint geometry labeling of integrated circuits (IC) is essential in defining the physical interface between components and the PCB layout, requiring precise visual perception. However, the unstructured nature of footprint drawings and abstract diagram annotations prevents direct IC footprint parsing and automated package geometry labeling methods from developing. Existing Large Multimodal Models (LMMs) struggle with inaccurate geometric perception, limiting their effectiveness in this task. To address these challenges, we propose LMM4-IC4K, a novel framework that treats IC mechanical drawings as images and leverages LMMs for structured geometric interpretation. To support such a framework, we introduce ICGeo8K, a multi-modal dataset with 8,608 labeled samples, including 4138 real-world IC footprint samples and 4470 synthetically generated samples. We further present a two-stage training framework to fine-tune LMMs for IC footprint labeling. Extensive experiments demonstrate that our model outperforms state-of-the-art LMMs on the proposed benchmark. The accurate translation of footprint diagrams enabled by LMM4-IC4K contributes to advancing automation and standardization within the PCB industry.
APA
Wang, Y., Lu, T., Liu, R., Yang, L., Chen, Z., Wang, Y., Liu, Y., Lin, K., Chen, X., Ding, D., Li, Y., Yang, Y., Chen, Y., Jin, Y. & Gowda, M.. (2026). LMM4-IC4K: A Large Multimodal Model Powered Integrated Circuit Footprint Geometry Understanding. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:128742-128761 Available from https://proceedings.mlr.press/v306/wang26fg.html.

Related Material