Zhihao Peng’s Academic Website
About Me
I am a graduate student at China University of Geosciences (Wuhan), currently pursuing a Master’s degree in Computer Science. My research focuses on Computer Vision, specifically in Single Image Super-Resolution (SISR). My work aims to improve the quality and efficiency of image reconstruction, particularly in low-resolution image enhancement.
With a strong background in deep learning, I am keen on developing lightweight and efficient architectures for real-time applications. My research interests lie in multi-scale feature fusion, global and local granularity optimization, and model optimization for better performance in practical scenarios.
Research Interests
- Single Image Super-Resolution (SISR)
- Deep Learning and Computer Vision
- Multi-Scale Feature Fusion and Lightweight Network Design
- Image Quality Enhancement and Loss Function Optimization
- Convolutional Neural Networks (CNNs) and Attention Mechanisms
Education
- China University of Geosciences (Wuhan) | Master’s Degree, Computer Science
September 2023 - Present
Supervisor: Professor Linquan Yang
Research Focus: Image Super-Resolution and Computer Vision
Academic Achievements
Papers
- “Lightweight Local and Global Granularity Selection Optimization Network (LGGSONet) for Single Image Super-Resolution”
- Abstract: This paper introduces a novel lightweight method for SISR, combining local and global granularity optimization to enhance image reconstruction quality while maintaining computational efficiency.
- Journal: Neural Networks (Under Review)
- “Improved YOLOv8-Based Multi-Scale Traffic Camera Image Detection Network”
- Abstract: This work proposes an improved YOLOv8 model for multi-scale vehicle detection in traffic monitoring systems, addressing challenges caused by varying vehicle scales and distances.
- Conference: PRICAI 2024
Conferences & Talks
- “Advancements in Efficient Image Super-Resolution Methods”
- Conference: PRICAI 2024
- Date: November 2024
Research Projects
1. LGGSONet: Lightweight Granularity Selection Optimization for Image Super-Resolution
Project Overview: This project focuses on the design of a lightweight super-resolution network that optimizes both local and global granularities to achieve superior image reconstruction while maintaining low computational cost.
2. Multi-Scale Traffic Camera Image Detection Network
Project Overview: We propose an improved multi-scale detection network built upon YOLOv8 to handle vehicle detection challenges in traffic monitoring caused by inconsistent vehicle scales due to distance and camera angle.
Technical Skills
- Programming Languages: Python, C++, MATLAB
- Deep Learning Frameworks: PyTorch, TensorFlow
- Computer Vision Tools: OpenCV, PIL, scikit-image
- Development Environment: Ubuntu, Docker, VMware
- Data Processing and Analysis: NumPy, Pandas, SciPy