Yaroslava Lochman

Master Student in Data Science and a Research Engineer at The Machine Learning Lab
Experienced in solving computer vision problems with classical methods (camera parameters estimation from single-view) and deep neural networks (object detection; semantic and instance segmentation for medical images).
Yaroslava is supervised by James Pritts.

Scholar · GitHub · LinkedIn · Kaggle

Ukrainian Catholic University, 2018 — 2020
MSc in Data Science

National Technical University of Ukraine “Kyiv Polytechnic Institute”, 2014 — 2018
BSc in System Analysis

The Machine Learning Lab, Ukrainian Catholic University
Research Engineer
Developed a new method for robust camera calibration from a single image. Contributed to the code for the joint undistortion and rectification of imaged repeated texture that is rigidly transformed on the scene plane within the robust estimation framework for rectification from coplanar repeated patterns. Captured and annotated an RGB-D dataset that is used to establish ground-truth metric-rectifying homographies for images with scene planes. Coded a labeling pipeline to partially automate the segmentation of scene planes.

Ciklum, R&D
Research Engineer
Developed a fruit detector for low-resolution drone images that works for highly-occluded scenes in which fruits are densely packed. Addressed dataset problems (duplicates, missing and false labels) with filtering, data distillation, and exhaustive train-time augmentation techniques. Developed a cell nucleus boundary detector for multi-modal images of cells based on an approach similar to Mask R-CNN. The method was robust to different cell types, magnification factors, and generalized well.

Deloitte, Technology Integration Department
IT Consultant
Automated human resource processes and financial statements analysis. Developed software to minimize time spent on tedious activities and human-factor risks.

Minimal Solvers for Single-View Auto-calibration.
Y. Lochman, J. Pritts, O. Dobosevych, R. Hryniv
( master thesis | code )

Minimal Solvers for Rectifying from Radially-Distorted Conjugate Translations.
J. Pritts, Z. Kukelova, V. Larsson, Y. Lochman, O. Chum.
arXiv preprint ( paper | code )

Minimal Solvers for Rectifying from Radially-Distorted Scales and Change of Scales.
J. Pritts, Z. Kukelova, V. Larsson, Y. Lochman, O. Chum.
arXiv preprint, accepted for publication in IJCV ( paper | code )

Geometry in Computer Vision, 2019  ( course notes)
Course Module for CS bachelors in UCU

Data Analysis and Preparation in Computer Vision, 2018  ( code | slides )
Lecture for CS bachelors in UCU

Segmentation of Microscopy Images, 2018  ( slides )
Lecture within the Deep Learning Talk, Women Who Code Kyiv