我是华北电力大学信息与通信工程专业的研究生，对计算机视觉和AI开发充满热情。我的研究重点是视觉语言模型（VLM）、视觉语言预训练（VLP）、目标检测和开放词汇检测（OVD）。
I'm a graduate student majoring in Information and Communication Engineering at
North China Electric Power University, with a strong passion for Computer Vision and AI
Development. My research focuses on Vision-Language Models (VLM), Vision-Language Pre-training
(VLP), Object Detection, and Open-Vocabulary Detection (OVD).
本科期间（2020-2024），我在通信原理、数字信号处理和计算机网络等方面打下了坚实的基础。现在作为硕士研究生（2024-2027），我正在深入探索前沿AI技术。我已获得华为HCIP-AI昇腾开发者认证，并持续探索计算机视觉领域的创新解决方案。
During my undergraduate studies (2020-2024), I built a solid foundation in
communication principles, digital signal processing, and computer networks. Now as a master's
student (2024-2027), I'm diving deeper into cutting-edge AI technologies. I'm certified as a
Huawei HCIP-AI Ascend Developer and continuously exploring innovative solutions in computer
vision.
我的研究生GPA为90.264，专业排名5/39。我致力于将理论知识与实践应用相结合，相信技术不仅应该突破边界，更应该解决现实世界的问题。
With a GPA of 90.264 and ranked 5/39 in my graduate program, I'm committed to
combining theoretical knowledge with practical implementation. I believe technology should not
only push boundaries but also solve real-world problems.
实习经历
Internship Experience
核电精密管螺纹缺陷检测研究
针对核电螺纹检测痛点，研发旋转驱动全景采集系统，攻克视觉盲区；改进YOLOv11引入纹理增强TFE模块，在小样本高反光环境下实现mAP提升至92.7%，单帧推理仅1.5ms。在大亚湾核电发表内刊论文《核电管接头螺纹缺陷视觉检测方法研究》。
Visual Defect Detection of Precision Pipe Threads
Developed
a rotation-driven panoramic acquisition system to overcome visual blind spots;
improved YOLOv11 with the TFE module, increasing mAP to 92.7% in high-reflection
environments with 1.5ms inference. Published an internal journal paper at Daya
Bay Nuclear Power Plant.
多端AI部署与云平台建设
主导多端AI部署架构。解决Android12+进程限制(Signal9)问题，基于Proot-Distro构建轻量化运行容器，实现手机端实时推理；利用TensorRT(FP16)在Jetson
Orin NX平台实现量化加速，封装一键启动工具与桌面快捷入口；搭建Argus云平台，通过Cloudflare &
Tailscale实现内网穿透与分布式组网支撑业务统一调用。
Multi-terminal AI Deployment & Cloud Platform
Resolved
Android 12+ Signal 9 issues using Proot-Distro for mobile inference; utilized
TensorRT (FP16) on Jetson Orin NX with visual startup tools; built the Argus
cloud platform with Cloudflare & Tailscale for distributed AI service
invocation.
综合工程实践与实验室管理
深度参与国家级技能大师工作室申报工作，负责论文专利等百余项科研成果的整理、归纳与材料编撰。熟练运用FDM3D打印技术实现结构件设计与制造；主导资产数字化建档与标签自动化处理，通过脚本工具实现50+项精密设备的数字化管理与规范化标识，提升实验室运维效率。
Engineering Practice & Lab Management
Participated in
National Skill Master Studio applications, compiling 100+ research achievements.
Utilized FDM 3D printing for structural design; led digital archiving and
automated labeling for 50+ precision devices, improving O&M efficiency.
专业技能
Professional Skills
视觉语言与目标检测模型
VLM & Object Detection
视觉语言模型 (VLM/VLP)：深入理解 CLIP，Yolo-world，Vision Transformer (ViT)
等模型的原理与应用，具备多模态数据预处理、模型微调和推理能力。
目标检测 (Object Detection)：熟练掌握 YOLO 系列 (v5/v8/v11)、Faster R-CNN、DETR
等主流检测算法。熟悉开放词汇检测 (Open-Vocabulary Detection, OVD) 理论，并具备丰富的课题实践经验。
VLM / VLP: Deep understanding of CLIP, YOLO-World, and Vision Transformer
(ViT) principles and applications, with expertise in multimodal data preprocessing,
model fine-tuning, and inference.
Object Detection: Proficient in mainstream detection algorithms including
YOLO series (v5/v8/v11), Faster R-CNN, and DETR. Familiar with Open-Vocabulary
Detection (OVD) theory with hands-on research project experience.
开发语言与深度学习框架
Languages & Frameworks
开发语言：主用 Python，熟悉 C/C++（主要用于模型部署与性能优化）。
主流框架：熟练掌握 PyTorch 及其生态圈（如 torchvision, torchaudio），并了解 TensorFlow 与 Mindspore
的基本应用。
Languages: Primarily Python, familiar with C/C++ specifically for edge
deployment and performance optimization.
Frameworks: Proficient in PyTorch and its ecosystem (torchvision,
torchaudio), with foundational knowledge of TensorFlow and Mindspore.
环境管理与版本控制
Environment & Version
Control
环境与容器化：熟悉 Linux/MacOS 系统操作；熟练运用 Docker 进行开发与部署环境的容器化构建；熟悉使用 Conda/venv 管理
Python 环境及使用 Tmux 等终端工具。
工程与协同：精通 Git 进行代码版本控制，熟悉 GitHub/GitLab、Lazygit 以及飞书工作流等研发协同模式。熟练使用 VS
Code、PyCharm 作为 IDE 进行项目开发，并使用 Jupyter Notebook/Lab 进行实验探究。
Environment: Familiar with Linux/macOS systems. Proficient in Docker for
containerized deployment setups, Conda/venv for Python environment isolation, and
terminal tools like Tmux.
Engineering & Collaboration: Master of Git version control, fluent in
GitHub/GitLab, Lazygit, and project workflows. Highly experienced with IDEs like VS
Code and PyCharm, leveraging Jupyter Notebook/Lab for experimental validation.
数据处理与可视化
Data & Visualization
数据处理：熟练使用 OpenCV、Pillow 等库进行图像数据增强与预处理工作；熟练掌握 Labelme、CVAT 等主流数据标注工具的运用。
结果可视化：掌握 Matplotlib 与 OpenCV (cv2) 可视化方案，能够高效进行图像渲染、数据分布与实验对比结果的可视化分析。
Data Processing: Skilled in image data augmentation and preprocessing using
OpenCV and Pillow; experienced with visual annotation tools like Labelme and
CVAT.
Visualization: Proficient in Matplotlib and OpenCV (cv2) for rendering image
outputs, statistical analysis charting, and evaluating comparative experimental
results.
