Berkeley, California, United States
• Built a visual log app to automate validating and improving the performance of Computer Vision algorithms for identifying over 1,000 products in retail environments (for a frictionless checkout company) using Streamlit, OpenCV, and Google Cloud
• Developed an image embedding explorer web app to analyze images from in-store cameras to identify similar products, and reduced image processing time by 50% with a caching feature
• Created a 98% accurate semantic segmentation model to detect occlusions of products, including pipelines to aggregate and preprocess images, using OpenCV and Gluon