Data for article: Neural Graph Mapping for Dense SLAM with Efficient Loop Closure

<p>Data to reproduce experiments in "Neural Graph Mapping for Dense SLAM with Efficient Loop Closure".</p> <p>It contains:</p> <ul> <li>a preprocessed version of the Kintinuous data from <a href="https://github.com/mp3guy/Kintinuous">here</a> (kintinuous_data.zip)</li> <li>novel sequences on the larger Replica scenes (replica_big_data.zip)</li> <li>ORB-SLAM2 results on <ul> <li>Kintinuous (kintinuous_slam.zip)</li> <li>Replica Big (replica_big_slam.zip)</li> <li>Replica iMAP (replica_slam.zip)</li> <li>NRGBD (nrgbd_slam.zip)</li> <li>ScanNet subset (scannet_subset_slam.zip)</li> </ul> </li> <li>Additional views for evaluation purposes as <a href="https://github.com/JingwenWang95/neural_slam_eval">proposed by Co-SLAM</a> for<br> <ul> <li>Replica Big (replica_big_data.zip)</li> <li>Replica iMAP (replica_coslam_eval.zip, modified from&nbsp;<a href="https://github.com/JingwenWang95/neural_slam_eval">here</a> with additional views for staircase and kitchen scenes, which were missing in the original data)</li> <li>NRGBD (nrgbd_coslam_eval.zip, from <a href="https://github.com/JingwenWang95/neural_slam_eval">here</a>)</li> </ul> </li> </ul>