Nancy Abdelkareem

Mass classification and segmentation from mammograms | Bachelor degree graduation project

• Developed a system that is able to classify the mass in the mammogram to benign or malignant, and to produce the segmentation of this mass using deep learning methods.
• Building an ensemble of deep neural networks able to produce the mass segmentation from the full mammogram image, and refine the segmentation with another ensemble trained on the mammogram’s region of interest (the mass)
• Building an ensemble of networks where each network is able to classify the mass to be benign or malignant and produce the result using majority voting.

DAVIS: Densely Annotated VIdeo Segmentation Supervised by Marwan Torki, Assistant Professor in the Computer and Systems Engineering, July. 2019 – Oct. 2019

• It’s about semi-supervised video object segmentation, the task of automatically generating accurate pixel masks for objects in a video sequence, given the first-frame ground truth annotations.
• Based on the concepts of Proposal-generation, Refinement, Optical Flow and Re-Identification.
• Tested on DAVIS data set using Google Colaboratory GPUs, Caffe framework.