poor result on new college dataset.

[Zhangjyhhh]

Hi, Thanks for your excellent work !!! I want to know how to get better result as your paper in TABLE XI on Quad. I use the default config file: ./config/Lidar_slam/run_ncd.yaml. i generate mesh use command " python vis_pinmap.py ./experiments/ncd**** 0.2 nueral_points.ply mesh_20cm.ply". and here is my result:

Predicted mesh unifrom sample: 10000000 --> 8563620 ( 0.02 m) {'MAE_accuracy (m)': 0.16234044696769342, 'MAE_completeness (m)': 0.1927699826745742, 'Chamfer_L1 (m)': 0.17755521482113382, 'Chamfer_L2 (m)': 0.23996385734393333, 'Precision [Accuracy] (%)': 64.42148915027659, 'Recall [Completeness] (%)': 62.22490167539342, 'F-score (%)': 63.304146324647974, 'Spacing (m)': 0.02, 'Inlier_threshold (m)': 0.2, 'Outlier_truncation_acc (m)': 0.4, 'Outlier_truncation_com (m)': 2.0}

could you please tell me how can get the same result as you. here are the mesh and my config file screenshot:

[YuePanEdward]

Hi, thanks for your interest in our work.

Here are the meshes I got for the NCD02_Quad mapping experiment. This folder includes the meshes of PIN-SLAM and two baseline methods (NKSR with KISS-ICP poses and SLAMesh). Numbers of the other four baseline methods are reported by Nerf-LOAM.

PIN-SLAM's evaluation results are:

MAE_accuracy (m),MAE_completeness (m),Chamfer_L1 (m),Chamfer_L2 (m),Precision [Accuracy] (%),Recall [Completeness] (%),F-score (%),Spacing (m),Inlier_threshold (m),Outlier_truncation_acc (m),Outlier_truncation_com (m)
0.11552514390375901,0.15247310990752128,0.13399912690564014,0.22192155677537126,83.47852386868307,80.73416443483003,82.08341197526036,0.02,0.2,0.4,2.0

We generally increase the iteration numbers for mapping, and increase the training data pool capacity. This would not affect the pose estimation accuracy but would slightly improve the reconstruction quality.

[Zhangjyhhh]

Hi, thanks for your interest in our work.

Here are the meshes I got for the NCD02_Quad mapping experiment. This folder includes the meshes of PIN-SLAM and two baseline methods (NKSR with KISS-ICP poses and SLAMesh). Numbers of the other four baseline methods are reported by Nerf-LOAM.

PIN-SLAM's evaluation results are:

MAE_accuracy (m),MAE_completeness (m),Chamfer_L1 (m),Chamfer_L2 (m),Precision [Accuracy] (%),Recall [Completeness] (%),F-score (%),Spacing (m),Inlier_threshold (m),Outlier_truncation_acc (m),Outlier_truncation_com (m)
0.11552514390375901,0.15247310990752128,0.13399912690564014,0.22192155677537126,83.47852386868307,80.73416443483003,82.08341197526036,0.02,0.2,0.4,2.0

We generally increase the iteration numbers for mapping, and increase the training data pool capacity. This would not affect the pose estimation accuracy but would slightly improve the reconstruction quality.

thanks for your quick reply

[sco09]

Hi, thanks for your interest in our work.

Here are the meshes I got for the NCD02_Quad mapping experiment. This folder includes the meshes of PIN-SLAM and two baseline methods (NKSR with KISS-ICP poses and SLAMesh). Numbers of the other four baseline methods are reported by Nerf-LOAM.

PIN-SLAM's evaluation results are:

MAE_accuracy (m),MAE_completeness (m),Chamfer_L1 (m),Chamfer_L2 (m),Precision [Accuracy] (%),Recall [Completeness] (%),F-score (%),Spacing (m),Inlier_threshold (m),Outlier_truncation_acc (m),Outlier_truncation_com (m)
0.11552514390375901,0.15247310990752128,0.13399912690564014,0.22192155677537126,83.47852386868307,80.73416443483003,82.08341197526036,0.02,0.2,0.4,2.0

We generally increase the iteration numbers for mapping, and increase the training data pool capacity. This would not affect the pose estimation accuracy but would slightly improve the reconstruction quality.

Hi! Thank you for this very interesting and solid work! I would like to follow up on this thread to ask if you could share this configuration parameter in the repository. I believe it would be very helpful for benchmarking. Thanks!