Key points are not available for this paper at this time.
Datasets have gained an enormous amount of popularity in the computer vision, from training and evaluation of Deep Learning-based methods to Simultaneous Localization and Mapping (SLAM). Without a doubt, imagery bears a vast potential due to scalability in terms of amounts data obtainable without tedious manual ground truth annotations or. Here, we present a dataset with the aim of providing a higher of photo-realism, larger scale, more variability as well as serving a range of purposes compared to existing datasets. Our dataset leverages availability of millions of professional interior designs and millions of-level furniture and object assets -- all coming with fine geometric and high-resolution texture. We render high-resolution and high-rate video sequences following realistic trajectories while supporting camera types as well as providing inertial measurements. Together with release of the dataset, we will make executable program of our interactive software as well as our renderer available at: //interiornetdataset. github. io. To showcase the usability and uniqueness our dataset, we show benchmarking results of both sparse and dense SLAM.
Li et al. (Mon,) studied this question.