CMDFE: A Robust Few-Shot Learning Framework with Tiny Object Feature Extraction for Environmental Microorganism Image Classification

Environmental Microorganisms (EMs) serve as vital bioindicators for monitoring environmental pollution monitoring. However, their classification is challenged by severe class imbalance and a lack of labeled data. Precise identification of EMs is essential for timely ecological risk assessment, yet conventional methods struggle in few-shot learning (FSL) contexts due to poor generalization and overfitting. To overcome these challenges, we introduce a novel Cross-stage Multi-branch Deep Feature Extraction (CMDFE) framework, which integrates deep feature extraction (Re) with a lightweight multi-branch module (C2f). This design improves feature diversity and mitigates overfitting, while maintaining a favorable balance between accuracy and computational efficiency. Experimental results on the EMDs7 dataset demonstrate the effectiveness of CMDFE, achieving <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{6 8. 4 0 \%}$</tex> accuracy in 1-shot and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{7 7. 0 0 \%}$</tex> in 5-shot tasks, surpassing most baseline methods. Additionally, CMDFE significantly reduces memory usage, making it well-suited for deployment in resource-constrained environments. Overall, CMDFE offers a robust and scalable approach for few-shot EM classification and holds promise for intelligent environmental monitoring applications.