中山大学研0
参与将数据交换格式从 JSON 改为 protobuf 的工作。在 X64 平台上测试显示，序列化性能约为 JSON 的 10倍，反序列化小对象时性能约为 JSON 的 2 倍，且序列化后的数据大小仅为 JSON 的一半，预计整体性能提升达 40%。 Participated in migrating data exchange format from JSON to protobuf. Tests on X64 platform showed serialization performance was about 10x faster than JSON, deserialization of small objects was 2x faster, and serialized data size was reduced by half, with an expected overall performance improvement of 40%.
开发并验证了一个深度学习模型，用于预测月龄后45周前早产儿视网膜病变(ROP)的发病概率和严重程度。在训练中，AUC分别为0.90（发病概率）和0.87（严重程度）；在测试中，AUC分别为0.94和0.88，召回率为100%，显示出很强的临床验证能力。 Developed and validated a deep learning model to predict the probability and severity of Retinopathy of Prematurity (ROP) before 45 weeks postmenstrual age. Achieved AUCs of 0.90 (occurrence) and 0.87 (severity) in training; 0.94 and 0.88 in testing, with 100% recall, demonstrating strong clinical validation capability.
通过搭载的 RealSense 深度相机精确获取对象位置信息，并运用 OpenCV 进行实时图像处理，显著提高了团队在场景分析和机器视觉方面的能力。测试场景中对目标识别准确率达到 99% 以上。 Utilized RealSense depth camera for precise object position detection and implemented real-time image processing with OpenCV, significantly enhancing the team's capabilities in scene analysis and computer vision. Achieved over 99% accuracy in target recognition during test scenarios.
GPA: 4.0
英语能力：CET-4、CET-6、雅思7分 English Proficiency: CET-4, CET-6, IELTS 7.0
我喜欢旅行，我相信，一个人行走的范围，就是他的世界。 I love traveling. I believe that the extent of one's travels defines the breadth of their world.