Development of DMS system using computer vision in ADAS team

CBG HardwareStack: Pytorch, OpenCV, Scikit-learn, Pandas, NumPy, ONNX Runtime.- Working on the development of a raw domain video enhancement model for joint learning of demosaicing, HDR fusion, quality enhancement and artifact removal.- Trained Unet-based models to solve image enhancement task for mobile devices with the help of GANs.- Experimented with various injections and augmentation techniques to optimize model performance.- Created a custom dataset preparation pipeline, streamlining preprocessing of images for model training.

Teaching Assistant in Computer Vision and Video Processing course

Course Assistant in Generative Models course for CS HSE

Stack: Python, PyTorch, COLMAP, Open3D, computer vision, git, Docker- Worked on the digitalization of State Historical Museum’s cultural heritage objects by obtaining 3D models with high-quality color from an accurate 3D scan with a mediocre color texture from 3D scanner and high-quality photos from DSLR cameras.- Performed several experiments for camera poses refinement task to acquire refined projected texture on 3D object’s mesh using PixSfM and PixLoc.

Stack: Python, PyTorch, CNN, computer vision, git, Docker, COLMAP, SQL, jupyter, PowerPoint- Published one article at IEEE international conferences.- Worked on the digitalization of State Historical Museum’s cultural heritage objects by obtaining 3D models with high-quality color from an accurate 3D scan with a mediocre color texture from 3D scanner and high-quality photos from DSLR cameras.- Performed several experiments for camera poses refinement task to acquire refined projected texture on 3D object’s mesh using PixSfM and PixLoc.- Developed image segmentation models for drone-based collision avoidance systems, including gathering and labeling a custom dataset to enhance model accuracy.

Stack: Python, openCV, PyTorch, computer vision, git, MediaPipe– Worked on developing CV system for mobile frameworks to track body posture for correctness.– Used in sports like golf, tennis etc or exercise tracking.– Worked with mediapipe and pose estimation models and converted to tf lite.

Developed a CTF Dashboard, a website for MEPHI in order to monitor and collect scientific works of undergraduate and graduate students, and a CTF Dashboard for MEPhI students using React, Redux, Material UI, webpack.

Developed a full stack web application, a multilingual platform for E-Commerce, using React and Redux, Semantic UI, webpack.