Rock image recognition method based on ECAResNet50 transfer learning
Intelligent rock recognition is crucial for geological research and engineering applications. In this paper, we propose an ECA-ResNet50 model that incorporates the ECA attention mechanism and migration learning. The model is validated and evaluated on a rock image dataset. Experiments show that the model achieves an accuracy of 95.4%, which significantly improves the rock image recognition capability.
Authors
- Haoyuan Lv
- Jinshan Sun (ORCID: https://orcid.org/0000-0003-3919-2153)
- Yù Zhang (ORCID: https://orcid.org/0000-0003-2100-621X)
- Hui Lan (ORCID: https://orcid.org/0000-0003-0020-0048)
- Wei Xue
Institutions
- China University of Geosciences (CN)
- Jianghan University (CN)
Publication Details
- Published
- 2025-03-21
- DOI
- https://doi.org/10.1109/iscait64916.2025.11010463
- Primary Topic
- Brain Tumor Detection and Classification
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Natural Science Foundation of Hubei Province