Generate Label Uncertainty with GLEnet: I reported an error when running step1.1:

[eaphan]

Generate Label Uncertainty with GLEnet: I reported an error when running step1.1:

[6Superman6]

Please make sure you execute step 1.1 successfully. You can check whether the checkpoints are well saved in each directory like fold_*.

Sorry, I didn't succeed in step 1.1 training just now. Now I can train normally. After the step 1.1 training is completed, I believe that step 1.2 should be able to run normally. @Eaphan

[6Superman6]

Thank you very much. Now step1.0-step1.3 can be successfully executed. @Eaphan

[Eaphan]

1. Do you have two GPUs on your machine? Note I use two GPUs for training the GLENet. If you only have only one GPU, you should change the command adaptively.
2. Have you installed the right version Pytorch? Can your train other models on your server?

[6Superman6]

My computer has two GPUs, which can train other models. There is also a missing file when executing step 1.1: cfgs/${exp_id}_ gen_ ori.yaml

[Eaphan]

Sorry, I have fixed the bug. The cfgs/exp20_gen.yaml should be cfgs/exp20_gen_ori.yaml.

I have just fixed some bugs and run off the process. You can pull the lastest change and try again. @6Superman6

[6Superman6]

Sorry, I have fixed the bug. The cfgs/exp20_gen.yaml should be cfgs/exp20_gen_ori.yaml.

I have just fixed some bugs and run off the process. You can pull the lastest change and try again. @6Superman6

Step 1.1 can now be executed successfully, but the following error is reported when executing step 1.2:

2023-02-17 16:31:31,845 INFO cfg.TAG: exp20_gen
2023-02-17 16:31:31,845 INFO cfg.EXP_GROUP_PATH:
2023-02-17 16:31:32,016 INFO Length of dense_gt_infos is 156
2023-02-17 16:31:32,016 INFO ### Aug params: flip=True, scale=[0.95, 1.05], rot=0.78539816,shift=0, enable_similar_type=True
Traceback (most recent call last):
File "test.py", line 203, in
main()
File "test.py", line 199, in main
eval_single_ckpt(model, test_loader, args, eval_output_dir, logger, epoch_id, dist_test=dist_test)
File "test.py", line 59, in eval_single_ckpt
model.load_params_from_file(filename=args.ckpt, logger=logger, to_cpu=dist_test)
File "/home/dell/project/pcdet/GLENet-main/cvae_uncertainty/model.py", line 371, in load_params_from_file
raise FileNotFoundError
FileNotFoundError

[Eaphan]

Please make sure you execute step 1.1 successfully. You can check whether the checkpoints are well saved in each directory like fold_*.

[mishimat34]

@Eaphan @6Superman6
I too have the same issues. Need your kind help.
Although I can use kitti_infos_train.pkl & kitti_dbinfos_train.pkl offered by @Eaphan but I want to study and tune model.py in cvae_uncertainty so that I must generate kitti_infos_train.pkl & kitti_dbinfos_train.pkl of my own.

1. I run ln -s data/kitti cvae_uncertainty, which doesn't work. So when running step 1.2: GLENet Prediction I have:

Traceback (most recent call last):
File "test.py", line 203, in
main()
File "test.py", line 189, in main
dist=dist_test, workers=args.workers, logger=logger, training=False
File "/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py", line 1047, in build_dataloader
logger=logger,
File "/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py", line 136, in init
with open(db_infos_path, 'rb') as f:
FileNotFoundError: [Errno 2] No such file or directory: 'kitti/kitti_dbinfos_train.pkl'

So I guessed and moved kitti/kitti_dbinfos_train.pkl into cvae_uncertainty/kitti (I built kitti under cvae_uncertainty).

2. Running step 1.2 again then I have:

Traceback (most recent call last):
File "test.py", line 203, in
main()
File "test.py", line 199, in main
eval_single_ckpt(model, test_loader, args, eval_output_dir, logger, epoch_id, dist_test=dist_test)
File "test.py", line 59, in eval_single_ckpt
model.load_params_from_file(filename=args.ckpt, logger=logger, to_cpu=dist_test)
File "/home/lab715/Code/GLENet/cvae_uncertainty/model.py", line 371, in load_params_from_file
raise FileNotFoundError

3. Seems I don't have training results in cvae_uncertainty/output/exp20_gen/fold_x/ckpt, so step 1.1 would be that for me:
   FOLD_IDX: 0
   FOLD_IDX: 1
   FOLD_IDX: 2
   FOLD_IDX: 3
   FOLD_IDX: 4
   FOLD_IDX: 5
   FOLD_IDX: 6
   FOLD_IDX: 7
   FOLD_IDX: 8
   FOLD_IDX: 9

How can I do to have the training results in fold_x/ckpt?
Thanks.

[Eaphan]

ln -s data/kitti cvae_uncertainty, which doesn't work

Why doesn't it work? Do you prepare the data of KITTI and create the directory well?

[lhsgwyz]

我也有同样的问题。需要你的帮助。虽然我可以使用kitti_infos_train.pkl & kitti_dbinfos_train.pkl提供的，但我想在cvae_uncertainty中学习和调整 model.py，以便我必须生成我自己的kitti_infos_train.pkl &kitti_dbinfos_train.pkl。

1. 我运行 ln -s data/kitti cvae_uncertainty，它不起作用。因此，在运行步骤 1.2：GLENet 预测时，我有：

回溯（最近一次调用）：文件“test.py”，第 203 行，在 main（） 文件“test.py”，第 189 行，在 main dist=dist_test，workers=args.workers，logger=logger，training=False 文件“/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py”，第 1047 行，在 build_dataloader logger=logger，文件“/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py”，第 136 行，init with open（db_infos_path， 'rb'） as f： FileNotFoundError： [Errno 2] 没有这样的文件或目录： 'kitti/kitti_dbinfos_train.pkl'

所以我猜测并将 kitti/kitti_dbinfos_train.pkl 移动到 cvae_uncertainty/kitti 中（我在 cvae_uncertainty 下构建了 kitti）。

2. 再次运行步骤 1.2，然后我有：

回溯（最近一次调用最后一次）：文件“test.py”，第 203 行，在 main（） 文件“test.py”，第 199 行，在 main eval_single_ckpt（model， test_loader， args， eval_output_dir， logger， epoch_id， dist_test=dist_test） 文件“test.py”，第 59 行，在 eval_single_ckpt model.load_params_from_file（filename=args.ckpt， logger=logger， to_cpu=dist_test）文件“/home/lab715/Code/GLENet/cvae_uncertainty/model.py”，第 371 行，load_params_from_file引发 FileNotFoundError

3. 似乎我没有 cvae_uncertainty/output/exp20_gen/fold_x/ckpt 的训练结果，所以步骤 1.1 对我来说就是这样： FOLD_IDX：0 FOLD_IDX：1 FOLD_IDX：2 FOLD_IDX：3 FOLD_IDX：4 FOLD_IDX：5 FOLD_IDX：6 FOLD_IDX：7 FOLD_IDX：8 FOLD_IDX：9

我怎样才能获得 fold_x/ckpt 的训练结果？ 谢谢。

@Eaphan @6Superman6 I too have the same issues. Need your kind help. Although I can use kitti_infos_train.pkl & kitti_dbinfos_train.pkl offered by @Eaphan but I want to study and tune model.py in cvae_uncertainty so that I must generate kitti_infos_train.pkl & kitti_dbinfos_train.pkl of my own.

1. I run ln -s data/kitti cvae_uncertainty, which doesn't work. So when running step 1.2: GLENet Prediction I have:

Traceback (most recent call last): File "test.py", line 203, in main() File "test.py", line 189, in main dist=dist_test, workers=args.workers, logger=logger, training=False File "/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py", line 1047, in build_dataloader logger=logger, File "/home/lab715/Code/GLENet/cvae_uncertainty/dataset.py", line 136, in init with open(db_infos_path, 'rb') as f: FileNotFoundError: [Errno 2] No such file or directory: 'kitti/kitti_dbinfos_train.pkl'

So I guessed and moved kitti/kitti_dbinfos_train.pkl into cvae_uncertainty/kitti (I built kitti under cvae_uncertainty).

2. Running step 1.2 again then I have:

Traceback (most recent call last): File "test.py", line 203, in main() File "test.py", line 199, in main eval_single_ckpt(model, test_loader, args, eval_output_dir, logger, epoch_id, dist_test=dist_test) File "test.py", line 59, in eval_single_ckpt model.load_params_from_file(filename=args.ckpt, logger=logger, to_cpu=dist_test) File "/home/lab715/Code/GLENet/cvae_uncertainty/model.py", line 371, in load_params_from_file raise FileNotFoundError

3. Seems I don't have training results in cvae_uncertainty/output/exp20_gen/fold_x/ckpt, so step 1.1 would be that for me:
   FOLD_IDX: 0
   FOLD_IDX: 1
   FOLD_IDX: 2
   FOLD_IDX: 3
   FOLD_IDX: 4
   FOLD_IDX: 5
   FOLD_IDX: 6
   FOLD_IDX: 7
   FOLD_IDX: 8
   FOLD_IDX: 9

How can I do to have the training results in fold_x/ckpt? Thanks.

I have the same issues as you, have you solved your problem yet?

[Eaphan]

You should complete step 1.1 before you start step 1.2.

[mishimat34]

I found that i can modify label training instruction from "for iter in `seq 0 9`" to "`seq 0 0`". I can train with only FOLD_IDX 0，but still can't generate the uncertainty label (still lacking evaluation data from FOLD_IDX 1 to 9) FOLD_IDX 0 and FOLD_IDX 9, are they corresponding to different parts of KITTI, respectively? Should I train the label uncertainty with entire FOLD_IDX 0 to 9? 我發現可以透過指令修改 for iter in `seq 0 9`至 `seq 0 0`，如此只做 FOLD_IDX: 0，一樣能訓練(但資料不足無法生成標籤)。 想請問 FOLD_IDX: 0到 FOLD_IDX: 9分別對應KITTI中不同的部分與樣本嗎？ 訓練標籤的時候一定要從FOLD_IDX 0 做到FOLD_IDX 9嗎? [image: Mailtrack] <https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality11&> Sender notified by Mailtrack <https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality11&> 24/04/09 下午08:05:16 YF Zhang ***@***.***> 於 2024年3月25日 週一 下午5:33寫道：

…

You should complete step 1.1 before you start step 1.2. — Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AZ35VPRRK7NLIXSIANUOSL3YZ7VQJAVCNFSM6AAAAAAU65EO36VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMJXGU3TOMJWGY> . You are receiving this because you commented.Message ID: ***@***.***>