Deep learning based camera pose estimation from monocular camera images has a recent uptake in Visual SLAM research . GODSAC* outperforms the state-of-the-artapproaches in pose estimation accuracy, as we demonstrate in our experiments . The approach augments posepredictions from a trained neural network with noisy odometry data through theoptimization of a pose graph . The results are based on a model of a human pose graph, rather than an analysis of individual features .

Author(s) : Alphonsus Adu-Bredu, Noah Del Coro, Tianyi Liu, Jingyu Song, Yuqing Zhang

Links : PDF - Abstract

Code :
Coursera

Keywords : pose - graph - godsac - based - camera -

Leave a Reply

Your email address will not be published. Required fields are marked *