Related Experiment Video
Updated: Jul 2, 2026
03:31
End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
Published on: December 15, 2023
A novel hybrid quantum dilated convolutional Kronecker network (QDCKN) for marine object detection and classification
N Kumaran1, Shyam Mohan J S2, Thamaraiselvi D3
1Department of IoT, Dhanalakshmi Srinivasan University, NH-45, Trichy Chennai Trunk Road, Samayapuram, (Near Samayapuram Toll Plaza), Tiruchirappalli, 621 112, Tamil Nadu, India.
Scientific Reports
|July 1, 2026
Summary
A new Quantum (Conceptual) Dilated Convolutional Kronecker Networks (QDCKN) framework improves underwater object detection. This method enhances accuracy and stability in challenging marine environments, outperforming existing models.
Area of Science:
- Marine Biology
- Computer Vision
- Image Processing
Background:
- Underwater object identification is difficult due to low light, turbidity, and complex backgrounds.
- Computer vision and object detection offer efficient solutions for marine biology applications.
Purpose of the Study:
- To introduce a novel hybrid framework, Quantum (Conceptual) Dilated Convolutional Kronecker Networks (QDCKN), for robust underwater object detection.
- To address limitations of current methods in handling non-uniform illumination, occlusions, and noise.
Main Methods:
- A two-stage framework integrating Savitzky-Golay filtering, YOLOv3 object detection, and a QDCNN-Fuzzy Deep Kronecker Network (DKN) module.
- Utilizes quantum-inspired dilated convolutions, fuzzy uncertainty modeling, and Kronecker-structured feature compression.
- Cohesive interaction between feature representation, uncertainty handling, and structured compression.
Main Results:
- QDCKN achieved 93.40% classification accuracy, 92.20% precision, and 94.60% recall on the Underwater Object Detection dataset.
- Outperformed established methods including SA-FPN, MarineDet, mResNet, and EDR.
- Demonstrated improved performance and stability under challenging underwater conditions.
Conclusions:
- The proposed QDCKN framework offers a significant advancement in underwater object detection and classification.
- The integrated approach effectively handles inherent challenges in underwater imaging.
- QDCKN provides a more robust and stable solution for marine biology applications.