Computer Vision Engineer
March 2020 to September 2020
Fort Washington, Pennsyvania, United States
- Analyzed image illumination data using PIL, Numpy and Pandas to remove illumination differences between different machines to increase portability across different bonder machines.
- Applied Linear regression using scikit-learn's Linear regression on optics movement data to ensure that the location of eyepoint corresponds well to the amount of motion
- Developed a 5-95 percentile thresholding algorithm for foreign material detection using OpenCV
- Implemented OpenCV's Back Projection algorithm for grayscale images for foreign material detection
- Improved upon the thresholded images using various OpenCV kernel sizes and morphological transformations
- Performed image augmentation using GIMP and ImageJ to test various foreign material detection algorithms
- Tested the developed algorithms against existing 5% thresholding and Blob detection algorithms resulting in a 90% improvement in foreign material detection
- Created and added 6 software tests using Robot testing framework and awk
- Performed image annotation for CNN development and training.
- Trained CNN on 1000+ images for anomaly detection using Tensorflow on Nvidia's K4000 Quadro series GPU
- Used Atlassian JIRA for agile software development workflow integrated with Bitbucket for version control and Confluence for software documentation.