Variational quantum enhanced deep transfer learning for small underwater aqua species image classification

Area of Science:

• Marine Biology
• Quantum Computing
• Artificial Intelligence

Background:

• Underwater species classification is vital for fisheries management and ecosystem monitoring.
• Challenges include poor image quality and high computational costs of current deep learning models.
• Existing methods struggle with underwater distortions like low visibility and lighting changes.

Purpose of the Study:

• To develop an efficient and accurate underwater species classification framework.
• To address the limitations of traditional deep learning models in marine environments.
• To integrate quantum computing with deep learning for improved feature representation.

Main Methods:

• Proposed a Lightweight Variational Quantum Enhanced Deep Transfer Learning framework.
• Integrated pre-trained classical convolutional neural networks with variational quantum circuits.
• Utilized quantum feature extraction to reduce computational complexity and enhance classification.

Main Results:

• Achieved high classification accuracy up to 99.25% on a small aquafarming species dataset.
• Demonstrated significantly fewer parameters and floating-point operations compared to traditional models.
• Ablation studies confirmed the positive impact of quantum layers on performance.

Conclusions:

• Quantum deep transfer learning offers a robust and efficient solution for underwater species classification.
• The proposed framework shows potential for resource-constrained applications in marine monitoring.
• This hybrid approach advances automated marine ecosystem analysis and sustainable fisheries management.