| layout | page |
|---|---|
| title | About |
Postdoctoral Researcher
Academy of Advanced Interdisciplinary Research (AAIR)
Guangzhou Institute of Technology (GIT)
Xidian University, Xi'an, China
I am a computer vision researcher with over six years of experience developing intelligent machine vision systems for industrial applications. My work focuses on building data-efficient deep learning solutions — especially weakly supervised, semi-supervised, and contrastive learning methods — to tackle real-world challenges in automated defect detection, anomaly localization, and quality inspection.
I am passionate about bridging the gap between cutting-edge AI research and practical industrial deployment.
- Weakly & Semi-Supervised Learning for Defect Segmentation
- Transformer-based Attention Mechanisms
- Pixel-level Contrastive Learning
- Industrial Anomaly Detection & Localization
- Vision-Language Models for Smart Manufacturing
- Scalable AI Systems for Real-World Deployment
As a Postdoctoral Researcher at Xidian University, I work on advancing machine vision technologies for smart manufacturing and intelligent inspection systems. My recent work includes STAC — a novel framework published in IEEE Transactions on Industrial Informatics (2026) — that significantly improves weakly supervised defect localization through saliency-guided transformer attention and pixel-level contrastive learning.
I believe the future of industrial AI lies in creating systems that are not only accurate and efficient, but also reliable and practical in real production environments.
I am always open to meaningful collaborations, discussions, and opportunities in industrial AI and computer vision.
📬 Get in touch
Email: djene.mengistu@gmail.com
GitHub: github.com/djene-mengistu