← В ленту
Регистрация: 28.03.2025

Marc Pereyra Marí

Специализация: Computer Vision
I have done research in academia in the field of image processing and computer vision. I have developed on of the fastest open-source PIV packages (quickPIV) and used it to automatically detected migration patterns during the embryonic development of the red flour beetle. I have also developed segmentation pipelines for biological data, both with traditional image processing techniques as well as with AI by training a maskRCCN model. Now, I am eager to work in industry. I would like to be part of a collaborative environment, and to contribute to the development of applications with tangible impacts.
I have done research in academia in the field of image processing and computer vision. I have developed on of the fastest open-source PIV packages (quickPIV) and used it to automatically detected migration patterns during the embryonic development of the red flour beetle. I have also developed segmentation pipelines for biological data, both with traditional image processing techniques as well as with AI by training a maskRCCN model. Now, I am eager to work in industry. I would like to be part of a collaborative environment, and to contribute to the development of applications with tangible impacts.

Скиллы

Computer Vision
Software development
Machine learning
Matlab
C
C++
MongoDB
SQlite
Javascript

Опыт работы

Research scientist in image processing
03.2021 - 09.2024 |Frankfurt Institute for Advanced studies
3D data, Julia, Python, Pytorch, Paraview, 3D Slicer
My role was developing software and image processing pipelines for analyzing 3D microscopy data. In particular, I specialized in quantifying cell movements in 3D and segmenting movement patterns of tissues. For this, I developed one of the fastest open-source PIV packages (quickPIV) in Julia. Optimizing PIV required delving to low-level implementation details of the FFT algorithm offered by the C library FFTW, as well as leveraging cuFFT to move computations to the GPU. In addition, I implemented a high-throughput automated pipelines for segmenting organoids based on a mask-RCNN model. I oversaw a group of students in annotating 3D datasets, and later trained the model, applied it, and analyzed the resulting 3D+t organoid predictions.

Образование

Physical biology (Магистр)
2017 - 2019
Goethe University Frankfurt
Genetics
2012 - 2016
Universitat Autonoma de Barcelona

Языки

АнглийскийПродвинутыйИспанскийРоднойНемецкийВыше среднегоПольскийБазовый