This is the official dataset from the paper 360DVO: Deep Visual Odometry for Monocular 360-Degree Camera, which is published on IEEE Robotics and Automation Letters (RA-L) 2026.
Project Page | Paper | Code
360DVO is a deep learning-based omnidirectional visual odometry (OVO) framework. It introduces a distortion-aware spherical feature extractor (DAS-Feat) and an omnidirectional differentiable bundle adjustment (ODBA) module. This repository provides the real-world… See the full description on the dataset page:
https://huggingface.co/datasets/chris1004336379/360DVO.
