自从我第一次接触高级模型以来，我就被它们的强大功能所吸引，这促使我在本科网络工程学习期间探索它们的原理。这让我发现了计算机视觉，尤其是在2021年GDG DevFest上见证了它在日常生活中的应用之后。视觉模型不仅仅是工具，它还能帮助人类完成诸如识别CT图像中的目标之类的任务。

为了准备研究生学习，我在本科阶段巩固了数学和计算机科学的基础，重点学习了算法、数据结构和网络安全等课程。我的网络工程专业让我磨练了实践技能，在学生信息管理数据库设计和Linux网络服务器安全配置等项目中取得了高分，巩固了我对操作系统和算法的了解。此外，我还学习了Coursera深度学习专项课程等在线课程，以加深对神经网络的理解。

除了课程学习之外，我还将所学知识应用于实际问题。 2020年，我加入了朱霞教授的项目，开发一个基于机器视觉的卡车盲区检测系统，选择YOLOv3和基于ResNet的网络对盲区中的物体进行多尺度检测。在研究了机器学习框架和相关文献后，我专注于编码、数据分析以及在Raspberry Pi 4B上部署系统。实时处理方面的挑战需要硬件升级和算法优化，我们通过集成英特尔NCS和Tiny-YOLO来提高速度。这个项目实现了90%的检测率，让我接触到了Python的深度学习框架，并激发了我对计算机视觉与深度学习协同作用的兴趣。

为了进一步积累经验，我在南京东南司法鉴定服务中心实习，担任数据研究员，与警方合作分析嫌疑人数据，协助开展了一项假冒商品的调查。后来，我在金陵科技学院技术服务中心有限公司参与了公交车定位系统的开发，编写脚本清理冗余数据，并与资深同事合作解决编码问题。这些任务丰富了我的应用计算机科学技能。

Since my initial encounter with advanced models, I have been fascinated by their power, which drove me to explore their principles through my undergraduate studies in network engineering. This led me to discover computer vision, especially after witnessing its applications in daily life at the 2021 GDG DevFest. Vision models go beyond tools, aiding humans in tasks like identifying targets in CT images.

To prepare for graduate studies, I strengthened my foundations in math and computer science as an undergraduate, focusing on courses like Algorithms, Data Structures, and Network Security. My network engineering major allowed me to hone practical skills, with high marks in projects like Student Information Management Database Design and Linux Network Server Security Configuration, reinforcing my knowledge of operating systems and algorithms. Additionally, I pursued online courses like Coursera’s Deep Learning Specialization to deepen my understanding of neural networks.

Beyond coursework, I applied my knowledge to real-world problems. In 2020, I joined Prof. Xia Zhu's project to develop a truck blind area detection system using machine vision, selecting YOLOv3 and ResNet-based networks for multi-scale detection of objects in blind spots. After studying machine learning frameworks and literature, I focused on coding, data analysis, and deploying the system on a Raspberry Pi 4B. Real-time processing challenges required hardware upgrades and algorithm optimization, which we addressed by integrating Intel NCS and Tiny-YOLO for increased speed. This project, achieving a 90% detection rate, introduced me to Python's deep learning frameworks and spurred my interest in the synergy between computer vision and deep learning.

Seeking further experience, I interned as a data researcher at Nanjing Southeast Forensic Science Service, where I helped analyze suspect data alongside police, aiding in an investigation of counterfeit goods. Later, at Jinling Institute of Technology Technical Service Center Ltd, I worked on a bus positioning system, writing scripts to clean redundant data and solving coding issues through collaboration with senior colleagues. These tasks enriched my applied computer science skills.