Sahar Yousefi
I am a computer vision engineer. My work currently focuses on transformer based key information extraction. Previously, I was a senior computer vision engineer at an AI-based startup which its objective was to automate the process of automative inspections. Former, I was a researcher at Leiden university medical center (LUMC) focusing on medical image processing projects.
Here are some of my recent projects:
- Donut-hole: an sparcified version of Donut model, achieved through model distillation, reduces the model density by 54% while preserving performance.
- fastMRI: organized by Facebook AI Research (FAIR) at NeurIPS2019, a research project to investigate the use of AI to make MRI scans up to 10X faster.
- DDAUnet: Dilated Dense Attention Unet for esophageal GTV segmentation.
- ASL-to-PET Translation: An CNN based model for translating ASL to PET images.
- Hadamard-te-ASL Reconstruction: Dynamic Angiography and Perfusion Reconstruction from Hadamard-te Arterial Spin Labeling of rank 8.
- Motion Detector: A novel motion detection method using 3D discrete wavelet transform.