Computer Vision Engineer

TL;DR

Computer Vision Engineer (AI): Enhancing a visual localization system for UAVs by transitioning from image retrieval to pose estimation and adding geometry calculations for the video layer. Focus on feature matching, metric offset computation, and real-time video-to-map overlay.

Локация: Можно удалённо (Москва)

Зарплата: до 230 000 ₽

Компания

MVP-версия дашборда для трекинга дрона в реальном времени.

Что делать

• Refactor mathematics to transition from retrieval to accurate Pose Estimation.
• Compute metric offset using SuperPoint + LightGlue feature matching, accounting for perspective distortions.
• Implement GPS correction by adding the computed metric offset to reference frame coordinates.
• Apply strict RANSAC filtering to eliminate outliers and provide a confidence indicator.
• Calculate precise GPS coordinates for the four corners of the current frame on the backend.
• Implement temporal smoothing using Kalman filter or exponential averaging to stabilize corner coordinates.

Требования

• Уровень Senior (от 5 лет в CV).
• Experience with Visual SLAM, 6DoF Pose Estimation, Visual Odometry.
• Proficient in OpenCV with a deep understanding of projection matrices, homography, and pixel-to-metric system coordinate translation.
• Experience optimizing pipelines for real-time performance (TensorRT/ONNX Runtime).