Related Experiment Video
Updated: May 16, 2025
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
8.9K
Application of open domain adaptive models in image annotation and classification
Sheng Li1, Zhousheng Chang2, Haizhen Liu3
1Innovation and Entrepreneurship Institute, Guangxi Normal University, Guilin, China.
Plos One
|May 14, 2025
Summary
This study introduces an adaptive model for open-set image annotation and classification, improving accuracy on unknown data. The method enhances generalization and robustness in computer vision tasks.
Area of Science:
- Computer Vision
- Machine Learning
Background:
- Image annotation and classification are vital for applications like medical imaging and surveillance.
- Existing methods struggle with unknown target domain data, leading to poor accuracy and generalization.
Purpose of the Study:
- To develop an adaptive image annotation classification model for open-set domains.
- To address domain distribution differences impacting classification performance.
Main Methods:
- Utilizes dynamic threshold control and subdomain alignment strategy.
- Incorporates a channel attention mechanism for feature extraction.
- Employs dynamic weight adjustment for cross-domain feature alignment.
Main Results:
- Achieved 93.5% accuracy on unknown target domains and 89.6% on known domains.
- Reached 89.6% check accuracy and 90.7% recall rate.
- Classification time is only 1.2 seconds.
Conclusions:
- The proposed model significantly improves accuracy and efficiency in open-set scenarios.
- Enhances robustness and generalization for image annotation and classification.
- Offers a novel approach to domain adaptation challenges in real-world applications.
