Vision-Based Ego-Positioning for Internet-of-Vehicle

This paper presents a method for ego-positioning with low cost monocular cameras for an IoV (Internet-of-Vehicle) system. To reduce the computational and memory requirements as well as the communication overheads, we formulate the model compression algorithm as a weighted k-cover problem for better preserving model structures. Specifically for real-world vision-based positioning applications, we consider the issues with large scene change and propose a model update algorithm to tackle these problems. A long-term positioning dataset with more than one month, 105 sessions, and 14,167 images is constructed. Based on both local and up-to-date models constructed in our approach, extensive experimental results show that sub-meter positioning accuracy can be achieved, which outperforms existing vision-based algorithms.