Object Detection Using Computer Vision for Medical Images
Successfully completed a proof of concept to democratize healthcare using computer vision.
My client wanted a proof of concept to understand if computer vision using deep learning could identify medical problems from images. I used convolutional neural networks (deep learning) to deliver a successful proof of concept for object detection on medical images with a small sample training dataset of 2,500 annotated images.
Tools used included Google Cloud Platform (GCP), Python, YOLO, and Jupyter notebook. Set up Ubuntu VM on GCP to use GPU, loaded necessary software, and trained the model using cleansed data. Analyzed results using object detection metrics. Ran another iteration with a larger training dataset and compared results to show improvement.
The client was pleased with the improvements shown and plans to obtain funding to scale the model to accommodate a larger training dataset.
A comprehensive report including an executive summary; approach; data cleansing procedure; GCP setup; training and testing procedure; object detection metrics and recommendations were submitted at the end of the PoC.
Computer Vision for Water Meter Images
Used computer vision and deep learning to "read" water meter images and determine the reading with an accuracy of 85%.
Planned and executed a project using computer vision and deep learning.
Read in over 36,000 images, converted them into tensors, matched them against readings from a CSV file and cleansed the data. Ensured that the images were of the correct data type and size. Shuffled image and reading data. Split into train, cross-validation (CV), and test loaded the InceptionV3 model and used its weights on train, CV, and test data for transfer learning. Created separate hdf5 files with features, labels and batches datasets for train, CV, and test batches. Trained on a shallow CNN model and transformed the output into the right dimensions for comparison. Plotted the loss and per digit accuracy. Calculated the accuracy for train, CV, and test data.
Iterated the above process with multiple hyper-parameters, sizes of network and also InceptionResNetV2 to improve the accuracy.
Utilized Floydhub Cloud Platform, Keras on TensorFlow, Python 3.x, Jupyter Notebook, and Excel to work through this project. Utilized an Agile Methodology to keep the project on track and provide visibility.
Worldwide Loyalty Management Solution
Let the implementation of a loyalty management solution in seven countries to reduce the cycle time from months to five days.
This project was implemented in seven countries simultaneously to strip-out a decades-old mostly manual business process that had unpredictable cycle times and poor customer service to a simpler, consistent mostly automated business process that cut the cycle time to five business days.