Article Details

Title · Authors · Abstract · Keywords · DOI · References

Facial Pore Segmentation Method Based on Lightweight Encoder and Feature Enhancement

YANG Zhou, WANG Jianchao

(School of Computer and Information Engineering, Shanghai Polytechnic University, Shanghai 201209, China)

Section：Computer Technology · Vol. 2026, Issue 3 · Pages：107-111 · 2096-4706（2026）03-0107-05 · CLC number：TP391.4 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2026.03.020

Abstract

To address challenges of structural ambiguity (similar to spots), sebum highlight interference, and multi-scale variations existing in facial pore segmentation, this paper proposes a method integrating a lightweight encoder and feature enhancement. This method employs a Depthwise Separable Convolution to construct a lightweight backbone network, thereby reducing complexity. The SE channel attention module is embedded to enhance pore feature response and suppress interference. It introduces depth information prediction as an auxiliary task, and enhances discriminative power by leveraging pore depression geometric features. Experiments on a self-built high-resolution facial pore dataset demonstrate that the method achieves an IoU of 0.872, Dice of 0.921, precision of 0.896, and recall of 0.913, significantly outperforming CNN and UNet baselines. It effectively reduces false positives of spots and false negatives of pores, providing an efficient and accurate segmentation tool for skin analysis and cosmetic evaluation.

Keywords

pore segmentation; Depthwise Separable Convolution; SE channel attention; depth information prediction; UNet

References

  1. [1] SABINA U，WHANGBO T K. Edge-based Effective Active Appearance Model for Real-time Wrinkle Detection [J].Skin Research and Technology，2021，27（3）：444-452.
  2. [2] ALYAMI J，REHMAN A，SADAD T，et al. Automatic Skin Lesions Detection from Images through Microscopic Hybrid Features Set and Machine Learning Classifiers [J].Microscopy Research and Technique，2022，85（11）：3600-3607.
  3. [3] TAI C A，JANES E，CZARNECKI C，et al. Double-condensing Attention Condenser: Leveraging Attention in Deep Learning to Detect Skin Cancer from Skin Lesion Images [J/OL].Sensors，2024，24（22）：7231[2025-06-12].https://doi.org/10.3390/s24227231.
  4. [4] GANTZ H Y，ZAMEZA P A，ZAINO M，et al. Validity of Facial Skin Analysis Pore Detection: A Comparative Analysis [J].Journal of Cosmetic Dermatology，2024，23（10）：3427-3431.
  5. [5] KIM H，KANG D，SEONG D，et al. Skin Pore Imaging Using Spectral-domain Optical Coherence Tomography: A Case Report [J].Biomedical Engineering Letters，2023，13（4）：729-737.
  6. [6] JO D J，SHIN J Y，NA S J. Evaluation of Changes for Sebum, Skin Pore, Texture, and Redness Before and After Sleep in Oily and Nonoily Skin [J].Skin Research and Technology，2022，28（6）：851-855.
  7. [7] TOLLENAERE M D，MEUNIER M，DURDURET A，et al. Facial Pore Refining by Targeting Dermal and Epidermal Functions: Assessment across Age and Gender [J].Journal of Cosmetic Dermatology，2024，23（10）：3395-3408.
  8. [8] PARVAR S Y，AMANI M，SHAFIEI M，et al. The Efficacy and Adverse Effects of Treatment Options for Facial Pores: A Review Article [J].Journal of Cosmetic Dermatology，2023，22（3）：763-775.
  9. [9] KIM B Y，CHOI J W，PARK K C，et al. Sebum, Acne, Skin Elasticity, and Gender Difference-which is the Major Influencing Factor for Facial Pores? [J].Skin Research and Technology，2013，19（1）：e45-e53.
  10. [10] ZUNAIR H，HAMZA B A. Sharp U-Net: Depthwise Convolutional Network for Biomedical Image Segmentation [J/OL].Computers in Biology and Medicine，2021，136：104699[2025-06-12].https://doi.org/10.1016/j.compbiomed.2021.104699.
  11. [11] CHEN L-C，PAPANDREOU G，KOKKINOS I，et al. DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs [J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2018，40（4）：834-848.
  12. [12] DISSANAYAKE B，MIYAMOTO K，PURWAR A，et al. New Image Analysis Tool for Facial Pore Characterization and Assessment [J].Skin Research and Technology，2019，25（5）：631-638.
  13. [13] SHUKLA M M，TRIPATHI B K，DWIVEDI T，et al. A Hybrid CNN with Transfer Learning for Skin Cancer Disease Detection [J].Medical & Biological Engineering & Computing，2024，62（10）：3057-3071.
  14. [14] CHIN C-L，YANG Z-Y，SU R-C，et al. A Facial Pore Aided Detection System Using CNN Deep Learning Algorithm [C]//2018 9th International Conference on Awareness Science and Technology (iCAST).Fukuoka：IEEE，2018：90-94.
  15. [15] YIN W，ZHOU D M，NIE R C. DI-UNet: Dual-branch Interactive U-Net for Skin Cancer Image Segmentation [J].Journal of Cancer Research and Clinical Oncology，2023，149（17）：15511-15524.

About the authors

Yang Zhou (April 2000-), male, Han ethnicity, from Ankang, Shaanxi Province, is currently pursuing a master's degree in image processing and machine vision.