Dec `18 – Jun `19
Geo data processing
- Images preprocessing; - Building instance segmentation and object detection models; - Optimizing models to improve performance.
Processing of images with trees fields to measure the vegetation index in dynamics. Creating a library to automatically align the rasters that were captured with the time difference. Creating and tuning Neural Network for tree crops segmentation on RGB and Multispectral data. Developing images alignment algorithm that uses neural network inside and results of raster segmentation to get alignment accuracy lower than 30 cm at images that were captured from 60-120m height.
Technologies Stack:OpenCV Python Python Numpy TensorFlow
A complex Data Science project
Jun `18 – Nov `18
Video analytics platform
ML/Computer vision engineer
- Changed tracking model to vanilla deep-sort; - Implemented multi camera tracking; - Improved time and performance of the model.
The scalable high performing platform empowered by some models and algorithms. It’s aimed to solve various business tasks with specific business patterns layer and deployment at distributed infrastructure.
Technologies Stack:OpenCV Python TensorFlow
Photo & Video
Data analytics system
V.N. Karazin KhNU