Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Deep Learning-Based Rock Particulate Classification Using Attention-Enhanced ConvNeXt

Created by
  • Haebom

Author

Anthony Amankwah, Chris Aldrich

Outline

In this paper, we propose an improved deep learning model based on ConvNeXt to improve the accuracy of rock size classification. The proposed model, CNSCA, improves upon the basic structure of ConvNeXt by adding self-attention and channel-attention mechanisms. The self-attention mechanism captures long-range spatial dependencies, while the channel-attention mechanism emphasizes information-rich feature channels, effectively capturing fine-grained local patterns and broad contextual relationships. We evaluate the model using a rock size classification dataset and compare it with three strong baseline models. Our results demonstrate that incorporating the attention mechanism significantly improves the model's performance on fine-grained classification tasks involving natural textures such as rocks.

Takeaways, Limitations

•
Takeaways:
◦
ConvNeXt-based CNSCA model outperforms existing models in rock size classification.
◦
Demonstrating the effectiveness of a hybrid design combining self-attention and channel-attention mechanisms.
◦
We present the potential of deep learning models to improve the performance of fine-grained classification tasks involving natural textures such as rocks.
•
Limitations:
◦
Lack of specific information about the dataset used (size, diversity, etc.).
◦
Lack of validation of generalization performance for different rock types or diverse environmental conditions.
◦
Lack of analysis of the computational cost and efficiency of the proposed model.
👍
Made with Slashpage