princeton-vl/DPVO

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1108 stars · 167 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DPVO is a deep learning-based visual odometry and SLAM system that estimates camera trajectories from video or image sequences using patch-based neural tracking. It implements the Deep Patch Visual Odometry (NeurIPS 2023) and Deep Patch Visual SLAM (ECCV 2024) papers from Princeton Vision Lab.

Use cases

• estimate camera trajectory from a video
• run visual SLAM on image sequences
• compute camera poses for 3D reconstruction
• track camera motion in monocular video
• benchmark visual odometry on datasets like TUM or EuRoC
• build AR or robotics localization pipelines

When to choose

• you need learning-based visual odometry with strong accuracy on monocular video
• you want a research-grade SLAM system with GPU acceleration and a real-time Pangolin viewer
• you need to reproduce or extend the DPVO/DPV-SLAM papers

When to avoid

• you need a CPU-only solution, since it requires CUDA GPUs
• you need a production-hardened SLAM stack with long-term support
• you need RGB-D or multi-camera sensor fusion out of the box

Facets

library · maturity active

computer-vision machine-learning deep-learning graphics computer-vision robotics autonomous-vehicles deep-learning python cpp visual-odometry slam camera-pose-estimation cuda research-code 3d-reconstruction linux gpu docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="princeton-vl/DPVO")