Yue Han Email & Phone Number

Mountain View, Ca, Us

San Francisco, California, United States

West Lafayette, Indiana, United States

• Research on image classification, detection, segmentation, object tracking, and synthetic image generation with deep learning• Focus on microscopy image analysis using machine learning and computer vision for segmentation and tracking• Developed a blur-image rejection system to increase detection results on egocentric image• Maintained and wrote analysis feature for the Android version of the Technology Assisted Dietary Assessment (TADA) app

West Lafayette, Indiana, United States

• Taught and delivered lectures in conditional distribution and Poisson Process• Participated in exam making and analyzed grades

Milpitas, California, United States

• Developed algorithms utilizing computer vision architectures such as 2D CNN or Vision Transformers to capture more domain information from spectra data, leading to a 10% improvement in predicting critical dimensions on wafers compared to the current method on RMSE• Proposed a novel embedding method that incorporates measurement specifications into any Transformer architecture, imbuing the models with physical meaning and resulting in an instant 2% performance boost in RMSE for improved prediction accuracy• Integrated the proposed architectures and embedding methods into the company’s research codebase, supporting the future development of metrology products

Palo Alto, California, United States

• Developed an algorithm that uses an ensemble of Mask R-CNN and Swin-Transformer to segment multi-channel fluorescence microscopy images and analyze the cells’ response under varying environmental conditions, increasing the precision and recall for nuclei segmentation by 9% and 10%, respectively• Collaborated with biologists and physicists on the capturing process and designed a GAN-based method to generate synthetic images of abnormally shaped nuclei with physical (e.g., location) and biological (e.g., shape) constraints• Modified the backbone of the CellPose segmentation network (U-Net) by adding an attention gate mechanism to the feature map, incorporating the environmental information into the network (e.g., correlation in locations of detected instances)• Created a cell response monitoring and analysis system consisting of individual segmentation modules to support the development of commercial products

Palo Alto, California, United States (Virtual)

• Researched on image-based small object tracking • Developed a tracking algorithm using the motion information from prior knowledge and it outperforms the current tracking method• Worked and cooperated with a team with diverse background knowledge

Wuhan, Hubei, China

Designed a new classifier for PolSar imagery which can tolerate the influence of speckle noise

Doctor Of Philosophy - Phd, Electrical Engineering

Bachelor Of Science - Bs, Electrical Engineering