Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization
This study presents a comprehensive comparison of U-Net and Ghost U-Net for road crack segmentation, emphasizing their performance and memory efficiency across various data representation formats, including FP32, FP16, and INT8 quantization. A dataset of 12,480 images was used, with preprocessing st...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ikatan Ahli Informatika Indonesia
2024-12-01
|
| Series: | Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) |
| Subjects: | |
| Online Access: | https://jurnal.iaii.or.id/index.php/RESTI/article/view/6089 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841544046407843840 |
|---|---|
| author | Haidhi Angkawijana Tedja Onno W. Purbo |
| author_facet | Haidhi Angkawijana Tedja Onno W. Purbo |
| author_sort | Haidhi Angkawijana Tedja |
| collection | DOAJ |
| description | This study presents a comprehensive comparison of U-Net and Ghost U-Net for road crack segmentation, emphasizing their performance and memory efficiency across various data representation formats, including FP32, FP16, and INT8 quantization. A dataset of 12,480 images was used, with preprocessing steps such as binarization and normalization to improve segmentation accuracy. Results show that Ghost U-Net achieved a marginally higher performance, with an IoU of 0.5041 and a Dice coefficient of 0.6664, compared to U-Net’s IoU of 0.5034 and Dice coefficient of 0.6662. Ghost U-Net also demonstrated significant memory efficiency, reducing GPU usage by up to 60% in FP16 and INT8 formats. However, a sharp decline in performance was observed for Ghost U-Net in the INT8 format, where the IoU dropped to 0.2038 and the Dice coefficient to 0.3227, whereas U-Net maintained stable performance across all formats. These findings suggest that Ghost U-Net is preferable for applications prioritizing memory efficiency and inference speed, while U-Net may be better suited for tasks requiring consistent accuracy across different quantization levels. This study underscores the importance of considering both performance stability and memory efficiency when selecting models for deployment in real-world applications. |
| format | Article |
| id | doaj-art-8f94bd3a9be34eda9e32af21c2390c69 |
| institution | Kabale University |
| issn | 2580-0760 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Ikatan Ahli Informatika Indonesia |
| record_format | Article |
| series | Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) |
| spelling | doaj-art-8f94bd3a9be34eda9e32af21c2390c692025-01-13T03:30:32ZengIkatan Ahli Informatika IndonesiaJurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)2580-07602024-12-018677978710.29207/resti.v8i6.60896089Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization OptimizationHaidhi Angkawijana Tedja0Onno W. Purbo1Institut Teknologi Tangerang SelatanInstitut Teknologi Tangerang SelatanThis study presents a comprehensive comparison of U-Net and Ghost U-Net for road crack segmentation, emphasizing their performance and memory efficiency across various data representation formats, including FP32, FP16, and INT8 quantization. A dataset of 12,480 images was used, with preprocessing steps such as binarization and normalization to improve segmentation accuracy. Results show that Ghost U-Net achieved a marginally higher performance, with an IoU of 0.5041 and a Dice coefficient of 0.6664, compared to U-Net’s IoU of 0.5034 and Dice coefficient of 0.6662. Ghost U-Net also demonstrated significant memory efficiency, reducing GPU usage by up to 60% in FP16 and INT8 formats. However, a sharp decline in performance was observed for Ghost U-Net in the INT8 format, where the IoU dropped to 0.2038 and the Dice coefficient to 0.3227, whereas U-Net maintained stable performance across all formats. These findings suggest that Ghost U-Net is preferable for applications prioritizing memory efficiency and inference speed, while U-Net may be better suited for tasks requiring consistent accuracy across different quantization levels. This study underscores the importance of considering both performance stability and memory efficiency when selecting models for deployment in real-world applications.https://jurnal.iaii.or.id/index.php/RESTI/article/view/6089ghost u-netimage segmentationmemory efficiencyquantizationu-net |
| spellingShingle | Haidhi Angkawijana Tedja Onno W. Purbo Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) ghost u-net image segmentation memory efficiency quantization u-net |
| title | Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization |
| title_full | Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization |
| title_fullStr | Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization |
| title_full_unstemmed | Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization |
| title_short | Performance and Efficiency Comparison of U-Net and Ghost U-Net in Road Crack Segmentation with Floating Point and Quantization Optimization |
| title_sort | performance and efficiency comparison of u net and ghost u net in road crack segmentation with floating point and quantization optimization |
| topic | ghost u-net image segmentation memory efficiency quantization u-net |
| url | https://jurnal.iaii.or.id/index.php/RESTI/article/view/6089 |
| work_keys_str_mv | AT haidhiangkawijanatedja performanceandefficiencycomparisonofunetandghostunetinroadcracksegmentationwithfloatingpointandquantizationoptimization AT onnowpurbo performanceandefficiencycomparisonofunetandghostunetinroadcracksegmentationwithfloatingpointandquantizationoptimization |