Headcam: Cylindrical Panoramic Video Dataset for Unsupervised Learning of Depth and Ego-Motion

This dataset contains panoramic video captured from a helmet-mounted camera while riding a bike through suburban Northern Virginia. We used the videos to evaluate an unsupervised learning method for depth and ego-motion estimation, as described in our paper: Alisha Sharma and Jonathan Ventura. "Unsupervised Learning of Depth and Ego-Motion from Cylindrical Panoramic Video." Proceedings of the 2019 IEEE Artificial Intelligence & Virtual Reality Conference, San Diego, CA, 2019. If you make use of this dataset, please cite this paper. The videos are stored as .mkv video files encoded using lossless H.264. To extract the images, we recommend using ffmpeg: mkdir 2018-10-03 ; ffmpeg -i 2018-10-03.mkv -q:v 1 2018-10-03/%05d.png ; Associated code can be found in our GitHub repository.

搜集汇总

背景与挑战

背景概述

Headcam数据集是一个用于深度和自运动估计无监督学习的圆柱形全景视频数据集，包含从头盔摄像头在郊区自行车骑行中捕获的视频，总大小为44.8 GB，以MKV格式存储，并附带相关论文和代码资源。