Related Experiment Video
Updated: Jul 20, 2025
04:48
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
442
SPA-Net: A Deep Learning Approach Enhanced Using a Span-Partial Structure and Attention Mechanism for Image Copy-Move
Kaiqi Zhao1, Xiaochen Yuan2, Zhiyao Xie2
1School of Computer Science and Engineering, Macau University of Science and Technology, Macao 999078, China.
Sensors (Basel, Switzerland)
|July 29, 2023
Summary
This study introduces SPA-Net, an efficient deep learning method for detecting image copy-move forgeries. SPA-Net accurately identifies manipulated regions by balancing copy-move and semantic features, outperforming existing methods.
Area of Science:
- Computer Vision
- Digital Image Processing
- Machine Learning
Background:
- Image copy-move forgery (CMFD) is a prevalent issue due to widespread visual sensor use and advanced digital image processing.
- Such forgeries, involving copying and pasting image regions, have significant implications in various sectors.
Purpose of the Study:
- To develop an efficient end-to-end deep learning approach for robust copy-move forgery detection.
- To enhance the accuracy of identifying manipulated image areas.
Main Methods:
- Proposed SPA-Net (span-partial structure and attention mechanism) for feature extraction.
- Utilized a span-partial structure to reduce redundant features and an attention mechanism to focus on tamper regions.
- Implemented a deep feature matching module for locating copy-move areas and a feature upsampling module for mask generation.
Main Results:
- SPA-Net achieved satisfactory performance on CASIA and CoMoFoD datasets.
- The proposed method demonstrated superior performance compared to existing CMFD techniques.
- Trained without pre-trained weights, SPA-Net effectively balanced copy-move and semantic features.
Conclusions:
- SPA-Net offers an efficient and effective solution for copy-move forgery detection.
- The developed SPANet-CMFD dataset aids in training robust forgery detection models.
- The approach shows potential for applications beyond CMFD, including deepfake detection.
