Viji Vennelakanti

Worked with a global team of data scientists, 基础架构工程师, 标签的公司, 报告作者, and researchers to build an innovative computer vision product. I was involved in all aspects of the project and wore multiple hats, including agile technical project manager, 产品经理, 业务分析师, 数据分析师, 和QA工程师.

主要成就:
- Worked with the founder to document requirements, 业务流程, 进入市场策略, 借助数据驱动的洞察力.
-帮助选择和管理与标签公司的关系以及他们所做的工作.
- Planned and implemented computer vision algorithms, training, and testing pipelines.
识别项目问题，并帮助团队集思广益解决问题.
- Devised the calculation of accuracy metrics at various stages of the pipeline.
-编写需求, managed, and tested an extensive rules-based algorithm to supplement computer vision.
- As the QA analyst, identified automation and outsourcing opportunities.
- Worked with a reporting company to produce complex custom reports.
- Managed the accounts, permissions, storage structure, and billing of AWS accounts.

我的客户想要一个POC来确定使用深度学习的计算机视觉是否可以从图像中识别医疗问题. 我使用卷积神经网络(深度学习)为医学图像上的目标检测提供了一个成功的POC，样本训练数据集为2,500张注释图像.

使用的工具包括Google Cloud Platform (GCP)、Python、YOLO和Jupyter笔记本. I set up Ubuntu VM on GCP to use GPU, 加载必要的软件, trained the model using cleansed data, and analyzed the results using object detection metrics. 然后我用一个更大的训练数据集进行了另一次迭代，并比较了结果以显示改进.

客户对所显示的改进感到满意，并计划获得资金来扩展模型以适应更大的训练数据集. I submitted a comprehensive report and recommendations at the end of the POC. The report included an executive summary, approach, 数据清理过程, GCP的设置, 培训和测试程序, 目标检测指标.

培训过程:
-阅读超过36000张图片.
-将它们转换成张量.
- Matched them against readings from a CSV file and cleansed the data.
- Ensured that the images were of the correct data type and size.
-打乱图像和读取数据.
- Split into train and cross-validation (CV).
—测试加载InceptionV3模型.
- Used its weights on train, CV, and test data for transfer learning.
- Created separate hdf5 files with features, labels, 还有火车的批处理数据集, CV, 测试批次.
-在一个浅CNN模型上训练，并将输出转换成正确的维度进行比较.
- Plotted the loss and per-digit accuracy.
- Calculated the accuracy for training, CV, and test data.

我用多个超参数和网络大小与InceptionResNetV2一起迭代了上述过程，以提高准确性，并使用敏捷方法使项目保持正轨并提供可见性. The technologies used to work through this project included Floydhub Cloud Platform, TensorFlow上的Keras, Python 3, Jupyter笔记本, and Excel.

This project was implemented in seven countries simultaneously to replace a decades-old, 大多是手工的业务流程，具有不可预测的周期时间和较差的客户服务, 一致的, and mostly automated business process that cut the cycle time to five business days.