Application of semi-supervised Mean Teacher to rock image segmentation
Accurate segmentation of rock images is crucial for studying the internal structure and properties of rocks. To address the issue of requiring a large number of labeled images for model training in traditional image segmentation methods, this paper proposes an improved semi-supervised Mean Teacher...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Slovenian Society for Stereology and Quantitative Image Analysis
2025-01-01
|
| Series: | Image Analysis and Stereology |
| Subjects: | |
| Online Access: | https://www.ias-iss.org/ojs/IAS/article/view/3279 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Accurate segmentation of rock images is crucial for studying the internal structure and properties of rocks. To address the issue of requiring a large number of labeled images for model training in traditional image segmentation methods, this paper proposes an improved semi-supervised Mean Teacher algorithm based on ResNet34-UNet. This method achieves relatively accurate rock image segmentation using only a small amount of labeled data. Initially, we use ResNet34-UNet as the base model to create Student Model and Teacher Model with identical structures. Then, we introduce self-attention mechanism into the semi-supervised Mean Teacher algorithm to further enhance its performance in rock image segmentation. Finally, by comparing the performance of supervised and semi-supervised Mean Teacher algorithms on image segmentation tasks, we validate the effectiveness of semi-supervised learning in rock image segmentation.
|
|---|---|
| ISSN: | 1580-3139 1854-5165 |