Hi！Thanks for the great work I'm running the previous version of the program.I can test successfully according to readme.md. first , Generate preprocessed data after running bash prepare_data.sh; second, I use your open source pre training model（softgroup_s3dis.pth）to test and test data is Area2. then run "python test_s3dis.py --config/softgroup_fold5_default_s3dis.yaml'', generate test results include semantic_pred、instance_pred; finally, I use script "visualization.py" for visualization。But, Inference result is much worse than ground true，I don't know where went wrong。I hope I can get your answer。

Hi, I got the following error when training on my own dataset. However, I could reproduce the results on S3DIS dataset without any error. Can you please look into my error and give me suggestion what might have been wrong? My dataset is same like S3DIS dataset with xyzrgb format. I have the following configuration: CUDA = V10.2.89 torch = 1.11.0 spconv-cu102 = 2.1.21 Ubuntu 18.04 I am really sorry for a long error message. Thank you for your help !

(softgroup2) shrijan_pf@BQ-DX1100-CT2:~/SoftGroup$ ./tools/dist_train.sh configs/softgroup_s3dis_backbone_fold5_mintA.yaml 1 2022-04-28 13:43:26,027 - INFO - Config: model: channels: 32 num_blocks: 7 semantic_classes: 3 #changed, original 13 instance_classes: 3 #changed, original 13 sem2ins_classes: [0, 1] semantic_only: True ignore_label: -100 grouping_cfg: score_thr: 0.2 radius: 0.04 mean_active: 300 class_numpoint_mean: [1823, 7457, 6189, 7424, 34229, 1724, 5439, 6016, 39796, 5279, 5092, 12210, 10225] npoint_thr: 0.05 # absolute if class_numpoint == -1, relative if class_numpoint != -1 ignore_classes: [-99] instance_voxel_cfg: scale: 50 spatial_shape: 20 train_cfg: max_proposal_num: 200 pos_iou_thr: 0.5 test_cfg: x4_split: True cls_score_thr: 0.001 mask_score_thr: -0.5 min_npoint: 100 fixed_modules: []

data: train: type: 's3dis' data_root: 'dataset/s3dis/preprocess_mint_sample' #changed original 'dataset/s3dis/preprocess' prefix: ['Area_2'] # changed original ['Area_1', 'Area_2', 'Area_3', 'Area_4', 'Area_6'] suffix: '_inst_nostuff.pth' repeat: 20 training: True voxel_cfg: scale: 50 spatial_shape: [128, 512] max_npoint: 2500000 min_npoint: 5000 test: type: 's3dis' data_root: 'dataset/s3dis/preprocess_mint_sample' #changed original 'dataset/s3dis/preprocess' prefix: 'Area_1' #changed, original 'Area_5' suffix: '_inst_nostuff.pth' training: False voxel_cfg: scale: 50 spatial_shape: [128, 512] max_npoint: 2500000 min_npoint: 5000

dataloader: train: batch_size: 1 #changed;original was 4 num_workers: 4 test: batch_size: 1 num_workers: 1

optimizer: type: 'Adam' lr: 0.004

save_cfg: semantic: True offset: True instance: False

fp16: False epochs: 20 step_epoch: 0 save_freq: 2 pretrain: './hais_ckpt_spconv2.pth' work_dir: ''

2022-04-28 13:43:26,027 - INFO - Distributed: True 2022-04-28 13:43:26,027 - INFO - Mix precision training: False 2022-04-28 13:43:27,398 - INFO - Load train dataset: 580 scans 2022-04-28 13:43:27,399 - INFO - Load test dataset: 32 scans 2022-04-28 13:43:27,399 - INFO - Load pretrain from ./hais_ckpt_spconv2.pth 2022-04-28 13:43:27,629 - INFO - removed keys in source state_dict due to size mismatch: semantic_linear.3.weight, semantic_linear.3.bias 2022-04-28 13:43:27,629 - INFO - missing keys in source state_dict: semantic_linear.3.weight, semantic_linear.3.bias 2022-04-28 13:43:27,629 - INFO - unexpected key in source state_dict: tiny_unet.blocks.block0.conv_branch.0.weight, tiny_unet.blocks.block0.conv_branch.0.bias, tiny_unet.blocks.block0.conv_branch.0.running_mean, tiny_unet.blocks.block0.conv_branch.0.running_var, tiny_unet.blocks.block0.conv_branch.0.num_batches_tracked, tiny_unet.blocks.block0.conv_branch.2.weight, tiny_unet.blocks.block0.conv_branch.3.weight, tiny_unet.blocks.block0.conv_branch.3.bias, tiny_unet.blocks.block0.conv_branch.3.running_mean, tiny_unet.blocks.block0.conv_branch.3.running_var, tiny_unet.blocks.block0.conv_branch.3.num_batches_tracked, tiny_unet.blocks.block0.conv_branch.5.weight, tiny_unet.blocks.block1.conv_branch.0.weight, tiny_unet.blocks.block1.conv_branch.0.bias, tiny_unet.blocks.block1.conv_branch.0.running_mean, tiny_unet.blocks.block1.conv_branch.0.running_var, tiny_unet.blocks.block1.conv_branch.0.num_batches_tracked, tiny_unet.blocks.block1.conv_branch.2.weight, tiny_unet.blocks.block1.conv_branch.3.weight, tiny_unet.blocks.block1.conv_branch.3.bias, tiny_unet.blocks.block1.conv_branch.3.running_mean, tiny_unet.blocks.block1.conv_branch.3.running_var, tiny_unet.blocks.block1.conv_branch.3.num_batches_tracked, tiny_unet.blocks.block1.conv_branch.5.weight, tiny_unet.conv.0.weight, tiny_unet.conv.0.bias, tiny_unet.conv.0.running_mean, tiny_unet.conv.0.running_var, tiny_unet.conv.0.num_batches_tracked, tiny_unet.conv.2.weight, tiny_unet.u.blocks.block0.conv_branch.0.weight, tiny_unet.u.blocks.block0.conv_branch.0.bias, tiny_unet.u.blocks.block0.conv_branch.0.running_mean, tiny_unet.u.blocks.block0.conv_branch.0.running_var, tiny_unet.u.blocks.block0.conv_branch.0.num_batches_tracked, tiny_unet.u.blocks.block0.conv_branch.2.weight, tiny_unet.u.blocks.block0.conv_branch.3.weight, tiny_unet.u.blocks.block0.conv_branch.3.bias, tiny_unet.u.blocks.block0.conv_branch.3.running_mean, tiny_unet.u.blocks.block0.conv_branch.3.running_var, tiny_unet.u.blocks.block0.conv_branch.3.num_batches_tracked, tiny_unet.u.blocks.block0.conv_branch.5.weight, tiny_unet.u.blocks.block1.conv_branch.0.weight, tiny_unet.u.blocks.block1.conv_branch.0.bias, tiny_unet.u.blocks.block1.conv_branch.0.running_mean, tiny_unet.u.blocks.block1.conv_branch.0.running_var, tiny_unet.u.blocks.block1.conv_branch.0.num_batches_tracked, tiny_unet.u.blocks.block1.conv_branch.2.weight, tiny_unet.u.blocks.block1.conv_branch.3.weight, tiny_unet.u.blocks.block1.conv_branch.3.bias, tiny_unet.u.blocks.block1.conv_branch.3.running_mean, tiny_unet.u.blocks.block1.conv_branch.3.running_var, tiny_unet.u.blocks.block1.conv_branch.3.num_batches_tracked, tiny_unet.u.blocks.block1.conv_branch.5.weight, tiny_unet.deconv.0.weight, tiny_unet.deconv.0.bias, tiny_unet.deconv.0.running_mean, tiny_unet.deconv.0.running_var, tiny_unet.deconv.0.num_batches_tracked, tiny_unet.deconv.2.weight, tiny_unet.blocks_tail.block0.i_branch.0.weight, tiny_unet.blocks_tail.block0.conv_branch.0.weight, tiny_unet.blocks_tail.block0.conv_branch.0.bias, tiny_unet.blocks_tail.block0.conv_branch.0.running_mean, tiny_unet.blocks_tail.block0.conv_branch.0.running_var, tiny_unet.blocks_tail.block0.conv_branch.0.num_batches_tracked, tiny_unet.blocks_tail.block0.conv_branch.2.weight, tiny_unet.blocks_tail.block0.conv_branch.3.weight, tiny_unet.blocks_tail.block0.conv_branch.3.bias, tiny_unet.blocks_tail.block0.conv_branch.3.running_mean, tiny_unet.blocks_tail.block0.conv_branch.3.running_var, tiny_unet.blocks_tail.block0.conv_branch.3.num_batches_tracked, tiny_unet.blocks_tail.block0.conv_branch.5.weight, tiny_unet.blocks_tail.block1.conv_branch.0.weight, tiny_unet.blocks_tail.block1.conv_branch.0.bias, tiny_unet.blocks_tail.block1.conv_branch.0.running_mean, tiny_unet.blocks_tail.block1.conv_branch.0.running_var, tiny_unet.blocks_tail.block1.conv_branch.0.num_batches_tracked, tiny_unet.blocks_tail.block1.conv_branch.2.weight, tiny_unet.blocks_tail.block1.conv_branch.3.weight, tiny_unet.blocks_tail.block1.conv_branch.3.bias, tiny_unet.blocks_tail.block1.conv_branch.3.running_mean, tiny_unet.blocks_tail.block1.conv_branch.3.running_var, tiny_unet.blocks_tail.block1.conv_branch.3.num_batches_tracked, tiny_unet.blocks_tail.block1.conv_branch.5.weight, tiny_unet_outputlayer.0.weight, tiny_unet_outputlayer.0.bias, tiny_unet_outputlayer.0.running_mean, tiny_unet_outputlayer.0.running_var, tiny_unet_outputlayer.0.num_batches_tracked, iou_score_linear.weight, iou_score_linear.bias, mask_linear.0.weight, mask_linear.0.bias, mask_linear.2.weight, mask_linear.2.bias 2022-04-28 13:43:27,630 - INFO - Training [Exception|implicit_gemm]feat=torch.Size([17843, 64]),w=torch.Size([32, 2, 2, 2, 64]),pair=torch.Size([8, 17848]),act=17848,issubm=False,istrain=True SPCONV_DEBUG_SAVE_PATH not found, you can specify SPCONV_DEBUG_SAVE_PATH as debug data save path to save debug data which can be attached in a issue. Traceback (most recent call last): File "./tools/train.py", line 185, in main() File "./tools/train.py", line 178, in main train(epoch, model, optimizer, scaler, train_loader, cfg, logger, writer) File "./tools/train.py", line 48, in train loss, log_vars = model(batch, return_loss=True) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/parallel/distributed.py", line 963, in forward output = self.module(*inputs[0], **kwargs[0]) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/shrijan_pf/SoftGroup/softgroup/model/softgroup.py", line 97, in forward return self.forward_train(**batch) File "/home/shrijan_pf/SoftGroup/softgroup/util/utils.py", line 171, in wrapper return func(*new_args, **new_kwargs) File "/home/shrijan_pf/SoftGroup/softgroup/model/softgroup.py", line 109, in forward_train semantic_scores, pt_offsets, output_feats = self.forward_backbone(input, v2p_map) File "/home/shrijan_pf/SoftGroup/softgroup/model/softgroup.py", line 263, in forward_backbone output = self.unet(output) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/shrijan_pf/SoftGroup/softgroup/model/blocks.py", line 139, in forward output_decoder = self.deconv(output_decoder) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/pytorch/modules.py", line 137, in forward input = module(input) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl return forward_call(*input, **kwargs) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/pytorch/conv.py", line 446, in forward input._timer, self.fp32_accum) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/cuda/amp/autocast_mode.py", line 118, in decorate_fwd return fwd(*args, kwargs) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/pytorch/functional.py", line 200, in forward raise e File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/pytorch/functional.py", line 191, in forward fp32_accum) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/pytorch/ops.py", line 1118, in implicit_gemm fp32_accum=fp32_accum) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/spconv/algo.py", line 661, in tune_and_cache GemmMainUnitTest.stream_synchronize(stream) RuntimeError: /io/build/temp.linux-x86_64-3.7/spconv/build/src/cumm/gemm/main/GemmMainUnitTest/GemmMainUnitTest_stream_synchronize.cc(11) CUDA error 700 terminate called after throwing an instance of 'c10::CUDAError' what(): CUDA error: an illegal memory access was encountered CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Exception raised from create_event_internal at /opt/conda/conda-bld/pytorch_1646755861072/work/c10/cuda/CUDACachingAllocator.cpp:1230 (most recent call first): frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x4d (0x7f5183c301bd in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libc10.so) frame #1: + 0x1ec97 (0x7f5183e9ac97 in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libc10_cuda.so) frame #2: c10::cuda::CUDACachingAllocator::raw_delete(void) + 0x23a (0x7f5183e9f1fa in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libc10_cuda.so) frame #3: + 0x2edaf8 (0x7f51ca243af8 in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libtorch_python.so) frame #4: c10::TensorImpl::release_resources() + 0x175 (0x7f5183c16fb5 in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libc10.so) frame #5: + 0x1daa99 (0x7f51ca130a99 in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libtorch_python.so) frame #6: + 0x4cf71c (0x7f51ca42571c in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libtorch_python.so) frame #7: THPVariable_subclass_dealloc(_object) + 0x299 (0x7f51ca425a39 in /home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/lib/libtorch_python.so) frame #8: + 0xfc359 (0x55c9017e7359 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #9: + 0xfac88 (0x55c9017e5c88 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #10: + 0xfac88 (0x55c9017e5c88 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #11: + 0xfa6a8 (0x55c9017e56a8 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #12: + 0xfb128 (0x55c9017e6128 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #13: + 0xfb13c (0x55c9017e613c in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #14: + 0xfb13c (0x55c9017e613c in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #15: + 0xfb13c (0x55c9017e613c in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #16: + 0xfb13c (0x55c9017e613c in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #17: + 0x12b4a7 (0x55c9018164a7 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #18: PyDict_SetItemString + 0x89 (0x55c901822a19 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #19: PyImport_Cleanup + 0xab (0x55c901897b8b in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #20: Py_FinalizeEx + 0x64 (0x55c90190c714 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #21: + 0x232e20 (0x55c90191de20 in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #22: _Py_UnixMain + 0x3c (0x55c90191e18c in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python) frame #23: __libc_start_main + 0xe7 (0x7f51e65b4c87 in /lib/x86_64-linux-gnu/libc.so.6) frame #24: + 0x1d803a (0x55c9018c303a in /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python)

ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: -6) local_rank: 0 (pid: 13703) of binary: /home/shrijan_pf/anaconda3/envs/softgroup2/bin/python Traceback (most recent call last): File "/home/shrijan_pf/anaconda3/envs/softgroup2/bin/torchrun", line 33, in sys.exit(load_entry_point('torch==1.11.0', 'console_scripts', 'torchrun')()) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 345, in wrapper return f(*args, **kwargs) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/distributed/run.py", line 724, in main run(args) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/distributed/run.py", line 718, in run )(*cmd_args) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/distributed/launcher/api.py", line 131, in call return launch_agent(self._config, self._entrypoint, list(args)) File "/home/shrijan_pf/anaconda3/envs/softgroup2/lib/python3.7/site-packages/torch/distributed/launcher/api.py", line 247, in launch_agent failures=result.failures, torch.distributed.elastic.multiprocessing.errors.ChildFailedError:

./tools/train.py FAILED

Failures: <NO_OTHER_FAILURES>

Root Cause (first observed failure): [0]: time : 2022-04-28_13:43:50 host : BQ-DX1100-CT2 rank : 0 (local_rank: 0) exitcode : -6 (pid: 13703) error_file: <N/A> traceback : Signal 6 (SIGABRT) received by PID 13703

When I am training on batch size 4 and numworkers 4 my training starts off at ~8gb. During the epochs it climbs up with steps of ~4gb. Eventually Ubuntu crashes, maybe because of another issue but the growing memory in the epoch is quite strange. I have not changed anything in the training or the softgroup model, is anyone aware of this issue and how can I fix the growing memory. I tried num workers 0 and increasing batch size but it keeps crashing, even in the first epoch compared to every ~ 4th epoch with batch 4 and num workers 4.

Hi,

Thanks for the open source code. As the title said, can I use the implementation on Windows? I see the installation instruction is binding to Linux system.

Many thanks, Enoch

Hi, thank you for your job! I made a dateset like S3DIS, but is smaller smaller than it. preprocess use --prepare_data.sh. But when I train, …… WRONG: point num < 20000, continue I try to change 20000 to 2000 or 20 , but it also happen. can you give me some advice when I use small data. THANK YOU!

I'm trying to train the model with a completely different dataset. I'm struggling to set the values in the config file, is there some documentation about setting them correctly or their meaning/influence?

Hi, when running visualize_open3d.py the parameter about the --prediction_path is passed in. /exp/scannetv2/softgroup/softgroup_default_scannet/result? Is this the result of scene0016_00?

Hello,when I run python train.py --config config/softgroup_fold5_default_s3dis.yaml,it throw the following error.Could you please tell me how to deal it?

Traceback (most recent call last): File "test_s3dis.py", line 338, in test(model, model_fn, data_name, cfg.test_epoch) File "test_s3dis.py", line 66, in test preds = model_fn(batch, model, epoch) File "/home/vipuser/PycharmProjects/pythonProject/SoftGroup_main1/model/softgroup/softgroup.py", line 482, in test_model_fn ret = model(input_, p2v_map, coords_float, batch_idxs, batch_offsets, epoch, 'test', split=True, semantic_only=semantic_only) File "/home/vipuser/miniconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py", line 547, in call result = self.forward(*input, **kwargs) File "/home/vipuser/PycharmProjects/pythonProject/SoftGroup_main1/model/softgroup/softgroup.py", line 385, in forward proposals_idx = torch.cat(proposals_idx_list, dim=0) RuntimeError: expected a non-empty list of Tensors

Hi, I encountered a suspicious problem after running a few hundreds of training epochs:

RuntimeError: DataLoader worker (pid 75890) is killed by signal: Killed.

The dataloader worker seemed to be stuck. Have you met the same problem and how can I solve it?

I have a dataset a point cloud. I labeled it using CloudCompare and add two Scalar Field values: the first is the semantic class and the second is the instance class. Could I train it on my dataset?

Hello, first of all, congratulation on your fantastic work. Currently, I have the code working. First, I prepared the dataset running the scripts:

• python split_data.py
• python prepare_data_inst.py --data_split val
• python prepare_data_inst.py --data_split test Once I had the dataset in an appropriated order, when I ran the latest version (https://github.com/thangvubk/SoftGroup/commit/cf88d9be41ae83a70f9100856a3ca15ee4ddcee9) of the inference (test.py) with the recommended configuration, I obtain semantic_prediction equal to 1 (in every point) on each point and the pred_instances is empty (in both cases: validation set using labels and test set without labeling). As an extra information, the only part which is different in my environment is the Pytorch version, which is 1.10.2. How can I solve this problem?

hello，author. There is such a function called 'ops.ballquery_batch_p'. May I ask what does it mean?The output is a number ( 66365386): nActive = ops.ballquery_batch_p(coords, batch_idxs, batch_offsets, idx, start_len, n, meanActive=300, radius=0.04)
if nActive <= n * meanActive(70329600): break meanActive = int(nActive // n + 1) idx = idx[:nActive] # (66365386): [10618,10632,...130336,230608] return idx, start_len

Hi! I trained SoftGroup/SoftGroup++ from scratch on STPLS3D, using configs: softgroup++_stpls3d.yaml and softgroup_stpls3d.yaml respectively. However, the training time in both cases is the same ~ 6 hours. Does SoftGroup also use Octree k-NN independent from "with_octree" config?

Hi,

Im training SoftGroup on my own dataset. I wondered whether ignore_label regards semantic labels or instance labels or both? I have labelled all points in my dataset that don't belong to any instance with -1 instance label. I therefore set ignore_label to -1.

Training the backbone on semantic segmentation worked fine, however, training the instance segmentation fails because of dimension errors:

/aten/src/ATen/native/cuda/Loss.cu:257: nll_loss_forward_reduce_cuda_kernel_2d: block: [0,0,0], thread: [8,0,0] Assertion t >= 0 && t < n_classes failed and CUDA error: device-side assert triggered

The error happens in instance_loss() calculation, more specifically line 232, in instance_loss cls_loss = F.cross_entropy(cls_scores, labels).

Could it be that the model does not ignore the -1 instances and therefore there is a dimension error?

Thank you in advance for any answers!

It seems like kitti config only existed in softgroup's config directory.

Do you have any plan for implementation of Softgroup++'s semantic kitti support?

Thanks

Hello! Thank you for your awesome work for instance segmentation. I notice that certain individual point may has multiple instance label by using Softgroup. PointGroup define the Instance segmentation task as finding the foreground objects in a scene and mark each object instance with a unique label. So, is multiple instance label appropriate?

When i train the model on ScannetV2, i confront the problem during validation "RuntimeError: CUDA error: an illegal memory access was encountered terminate called after throwing an instance of 'c10::CUDAError' what(): CUDA error: an illegal memory access was encountered"

So, is it concerned about the memory size? and if i skip validation， how can i deal with the problem during test ?