Dear authors, I notice that adaptive_feature = 'xyz' is selected for training, but according to you paper, it is adaptive_feature = 'xyz2_feat2' that should be used. By the way, it seems that adaptive_feature = 'asmy' never be used, because you chose if self.feat_mode == 'asym': rather than if 'asym' in self.feat_mode: in the code, and I think adaptive_feature = 'asmy' can't be use for its own. Maybe I do not get your idea... Can you please evaluate whether what I said is correct?

我注意到你们选择了 adaptive_feature = 'xyz' 进行训练，但根据你的论文，应该使用的是 adaptive_feature = 'xyz2_feat2'。还有个小问题，似乎从未使用过 adaptive_feature = 'asmy'，因为您在代码中选择了 if self.feat_mode == 'asym': 而不是 if 'asym' in self.feat_mode:，因此我觉得 adaptive_feature = 'asmy' 不能单独使用（xyz和feat2就表示不了了）。也许我没很好地理解你们的想法......你能评估一下我说的是否正确吗？

Thanks.

Regards

Dear Haoran,

Thanks for such excellent work.

I got some questions about S3DIS semantic segmentation task.

1. How many epochs do we need to train the model for the S3DIS semantic segmentation task?
2. Can you please give me some suggestions for the model training on S3DIS?
3. If it is possible that you share the trained model for S3DIS semantic segmentation task?

I would be very grateful for your help and hope you have a nice day.

Best Wishes, Haoran

Many of the datasets in every Area files cannot be transformed, not only the previously mentioned one "Area_5/hallway_6/Annotations"! Can anyone give some advice?

Dear author,

Thanks for your such excellent work.I'm going to use the architectures of your model for scene segmentation on our data set.

I got some questions about own data sets semantic segmentation task， When the model is being trained, it always gets the “Floating point exception(core dumped) ”and exits.

I hope the author can help me solve this problem.

e001-i0405 => L=0.507 acc= 77% / t(ms): 174.8 206.6 652.3) e001-i0415 => L=0.172 acc= 94% / t(ms): 175.0 197.6 634.6) e001-i0424 => L=0.280 acc= 88% / t(ms): 178.0 212.6 663.4) e001-i0433 => L=0.243 acc= 89% / t(ms): 184.6 238.5 742.1) e001-i0444 => L=0.508 acc= 81% / t(ms): 169.0 214.4 669.0) e001-i0455 => L=0.155 acc= 98% / t(ms): 167.2 200.8 623.6) e001-i0465 => L=0.289 acc= 87% / t(ms): 174.1 184.1 700.1) e001-i0476 => L=0.252 acc= 92% / t(ms): 180.9 223.2 689.1) e001-i0486 => L=0.187 acc= 94% / t(ms): 183.3 229.4 702.4) e001-i0496 => L=0.408 acc= 84% / t(ms): 177.2 194.4 725.5) Validation : 6.0% (timings : 93.59 91.53) Validation : 12.5% (timings : 154.77 138.18) Validation : 19.5% (timings : 167.47 157.46) Validation : 26.5% (timings : 173.43 170.19) Validation : 33.0% (timings : 180.49 182.67) Validation : 40.0% (timings : 188.73 183.80) Validation : 46.0% (timings : 199.89 195.97) Validation : 53.0% (timings : 193.20 194.36) Validation : 59.5% (timings : 196.38 194.34) Validation : 67.0% (timings : 190.50 174.29) Validation : 73.5% (timings : 189.48 186.74) Validation : 80.5% (timings : 184.58 190.03) Floating point exception(core dumped)

Best Wishes, q-i-zhang

@hrzhou2 尊敬的作者您好，我跑了一下您提供的S3DIS数据集的pretrained model，在test上miou大概是59.99，低于您文章中的精度。但是我在您的模型上作出了修改，性能有一定的提升，想投文章，所以期待得到更好的结果，希望和您取得联系可以吗？我的qq：1253174115（不好意思网上提供微信不太安全），来自哈工大，期待您的回复！谢谢！ @hrzhou2

Hello, dear author, after I downloaded the S3DIS data set, I encountered the following error when I was executing python train.py. I hope the author can help me solve this problem. 你好，亲爱的作者，我下载好S3DIS数据集后，当我在执行到python train.py的时候，我遇到了以下错误，希望作者能帮助我解决这个问题。 Cloud Area_4 - Room 21/49 : storage_4 Cloud Area_4 - Room 22/49 : hallway_13 Cloud Area_4 - Room 23/49 : storage_3 Cloud Area_4 - Room 24/49 : office_9 /home/lil/AdaptConv-master-main/sem_seg/datasets/S3DIS.py:678: UserWarning: loadtxt: Empty input file: "./data/Stanford3dDataset_v1.2/Area_4/office_9/Annotations/chair_3.txt" object_data = np.loadtxt(object_file, dtype=np.float32) Traceback (most recent call last): File "/home/lil/AdaptConv-master-main/sem_seg/train.py", line 766, in main() File "/home/lil/AdaptConv-master-main/sem_seg/train.py", line 226, in main training_dataset = S3DISDataset(config, args.dataset, set='training', use_potentials=True) File "/home/lil/AdaptConv-master-main/sem_seg/datasets/S3DIS.py", line 133, in init self.prepare_S3DIS_ply() File "/home/lil/AdaptConv-master-main/sem_seg/datasets/S3DIS.py", line 681, in prepare_S3DIS_ply cloud_points = np.vstack((cloud_points, object_data[:, 0:3].astype(np.float32))) IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed

Thanks