Mikhail Artemyev

About

Mikhail Artemyev is a computer vision researcher whose work centers on single-view depth estimation (SVDE), 3D geometry, and robust training methodologies for neural networks. His most influential contribution, the 2022 paper "Single-Stage 3D Geometry-Preserving Depth Estimation Model Training on Dataset Mixtures with Uncalibrated Stereo Data," addresses a critical bottleneck in SVDE: the reliance on diverse, high-volume training data. Artemyev proposed a novel single-stage framework that preserves 3D geometric consistency while training on mixed datasets with uncalibrated stereo inputs—a practical solution that avoids the need for expensive, perfectly calibrated data. This work has garnered 5 citations, signaling its growing relevance in robotics, augmented reality, and 3D modeling communities. By enabling more accurate depth estimation from a single RGB image, Artemyev’s research directly supports applications where scene geometry is paramount, such as autonomous navigation and virtual environment reconstruction. His approach stands out for its elegance in handling dataset heterogeneity, a common real-world challenge. For students and researchers entering computer vision, Artemyev’s work exemplifies how thoughtful model design can overcome data limitations, pushing the boundaries of what’s achievable in geometry-preserving depth estimation.

Research Focus

Key Achievements

Top Papers

1. 1
   Single-Stage 3D Geometry-Preserving Depth Estimation Model Training on Dataset Mixtures with Uncalibrated Stereo Data

   5 citations · 2022

Key Collaborators

Contact & Links