LayerMask: PersonMaskUltra和LayerMask: PersonMaskUltra V2

[liulsg]

问题：不能正常运行。 已经运行过repair_dependency.bat 这两个命令在我测试的过程中一直不能运行。下面是错误日志

Error occurred when executing LayerMask: PersonMaskUltra:

[Errno 13] Permission denied: 'G:\AI\ComfyUI_M\ComfyUI\models\mediapipe\selfie_multiclass_256x256.tflite'

File "G:\AI\ComfyUI_M\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "G:\AI\ComfyUI_M\ComfyUI\execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "G:\AI\ComfyUI_M\ComfyUI\execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "G:\AI\ComfyUI_M\ComfyUI\custom_nodes\ComfyUI_LayerStyle\py\person_mask_Ultra.py", line 63, in person_mask_ultra with open(a_person_mask_generator_model_path, "rb") as f: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[chflame163]

没用repair_dependency.bat修复之前也是这样吗？错误信息显示是文件权限问题不能打开，请查看你的protobuf依赖包版本

[liulsg]

已经安装到了最新版本，protobuf==5.26.0 我用你的.bat安装试过不可以。我自己手动安装仍然不能成功运行。 修复之前也是不可以运行的，

[kakachiex2]

I have the same error I install dependencies and the errors keep pop

Error occurred when executing LayerMask: PersonMaskUltra V2:

[Errno 13] Permission denied: 'K:\ComfyUI\ComfyUI\models\mediapipe\selfie_multiclass_256x256.tflite'

File "K:\ComfyUI\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "K:\ComfyUI\ComfyUI\execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "K:\ComfyUI\ComfyUI\execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "K:\ComfyUI\ComfyUI\custom_nodes\ComfyUI_LayerStyle\py\person_mask_ultra_v2.py", line 69, in person_mask_ultra_v2 with open(a_person_mask_generator_model_path, "rb") as f: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[kakachiex2]

@liulsg Try this [ pip install protobuf==4.25.3 ] it will uninstall the current version and install this one work for me but dont fix the errors.

[kakachiex2]

Deleting the folder mediapipe and running the node will create a new one and download the model this fix the problem for me.

[liulsg]

经过测试，出现此问题的原因是：自动下载的模型锁了权限 解决办法：删除 mediapipe 文件，手动创建mediapipe 文件并到官网下载模型放入这个文件夹内（与protobuf版本关系不大，我测试升降版本都没有问题），下面是原网址解决办法和模型下载地址。 https://github.com/djbielejeski/a-person-mask-generator/issues/28 https://developers.google.com/mediapipe/solutions/vision/image_segmenter#multiclass-model

---

After testing, the reason for this problem is: the automatically downloaded model locks permissions Solution: delete the mediapipe file, manually create the mediapipe file and download the model to the official website and put it in this folder (with protobuf version has little to do with it, I tested the upgrade version without problems), the following is the original URL solution and model download address.

[liulsg]

不过这个节点的判断逻辑应该还有优化的空间。 检测不到对应信息直接报错（比如开启检测衣服，未检测到直接报错）

[the2ndface]

这是哪个节点，求链接

[dabiaog]

这个节点的判断逻辑应该还有优化的空间。 检测不到对应信息直接报错（比如开启检测衣服，未检测到直接报错）

[T1987125630]

这个问题非常好解决，selfie_multiclass_256x256.tflite这个文件下载时生成了一个同名文件夹，把这个文件从文件夹里拿出来就OK了。 This problem is very easy to solve, selfie_multiclass_256x256.tflite this file download generated a folder with the same name, the file out of the folder is OK.

[yxf551]

Error occurred when executing LayerMask: PersonMaskUltra V2:

We couldn't connect to 'https://huggingface.co' to load this file, couldn't find it in the cached files and it looks like hustvl/vitmatte-small-composition-1k is not the path to a directory containing a file named config.json. Checkout your internet connection or see how to run the library in offline mode at 'https://huggingface.co/docs/transformers/installation#offline-mode'.

[3105083684]

执行 LayerMask 时出错：PersonMaskUltra V2：

我们无法连接到“https://huggingface.co”来加载此文件，在缓存文件中找不到它，并且看起来 hustvl/vitmatte-small-composition-1k 不是包含名为 config.json 的文件的目录的路径。在“https://huggingface.co/docs/transformers/installation#offline-mode”中签出您的互联网连接或查看如何在离线模式下运行库。

+1

[xjtuwolf]

Error occurred when executing LayerMask: SegmentAnythingUltra V2:

('Connection broken: IncompleteRead(149015986 bytes read, 175360628 more expected)', IncompleteRead(149015986 bytes read, 175360628 more expected))

File "D:\ai\ComfyUI\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all)

[H-pony]

Error occurred when executing LayerMask: PersonMaskUltra V2:

Failed to import transformers.models.pvt_v2.configuration_pvt_v2 because of the following error (look up to see its traceback): No module named 'transformers.models.pvt_v2.configuration_pvt_v2'

"How can this be solved, big shots?"

[chflame163]

Failed to import transformers.models.pvt_v2.configuration_pvt_v2 because of the following error (look up to see its traceback): No module named 'transformers.models.pvt_v2.configuration_pvt_v2'

Try update the transformers dependency

[plasx520]

Have yours been solved? I'm from https://[developers.google.com](https://developers.google.com/mediapipe/solutions/vision/image_segmenter#multiclass-model)/mediapipe/solutions/vision/image_segmenter#multiclass-model https://hf-mirror.com/yolain/selfie_multiclass_256x256/tree/main Whether creating files manually or automatically has all been tried or Error occurred when executing LayerMask: PersonMaskUltra V2: [Errno 13] Permission denied: 'H:\\sd\\ComfyUI\\models\\mediapipe\\selfie_multiclass_256x256.tflite'

[trustmeimanidiot]

I am still getting this error. Downloading the file manually and putting it in a manually created \ComfyUI\models\mediapipe\selfie_multiclass_256x256.tflite directory results in the same error.

Deleting the mediapipe folder entirely and letting Comfy recreate it also gives the same error.

I am using the Windows portable standalone install.

[vootox]

Have yours been solved? I'm from https:// developers.google.com /mediapipe/solutions/vision/image_segmenter#multiclass-model https://hf-mirror.com/yolain/selfie_multiclass_256x256/tree/main Whether creating files manually or automatically has all been tried or Error occurred when executing LayerMask: PersonMaskUltra V2: [Errno 13] Permission denied: 'H:\\sd\\ComfyUI\\models\\mediapipe\\selfie_multiclass_256x256.tflite'

I'm getting the same error

[plasx520]

感谢投稿怕冷爱上雪，您的投稿已收到，当晚就会开始审核。分享精神最可贵，如果稿件没有通过也非常感谢您投稿！软件无怕冷爱上雪版权的审核通过也不会发放奖励请勿使用快压流氓软件打包压缩包 发现一律略过

[offmybach]

I am still getting this error. Downloading the file manually and putting it in a manually created \ComfyUI\models\mediapipe\selfie_multiclass_256x256.tflite directory results in the same error.

Deleting the mediapipe folder entirely and letting Comfy recreate it also gives the same error.

I am using the Windows portable standalone install.

same here

[neom315]

error still persist, no solution found.

i have delete the folder and have comfy to create a new one --------- not working i have made the folder manually and download the model-------------not working i have moved the selfie_multiclass_256x256.tflite up one level as instructed in a chinese website and i get this error: Incorrect path_or_model_id: 'E:\Comfy--UI\ComfyUI_windows_portable\ComfyUI\models\vitmatte'. Please provide either the path to a local folder or the repo_id of a model on the Hub

i have lower the comfyui manager to weak and low------------------not working

at this point i don't see how this error could be solved.

[VitoMao]

The error indicates a permission issue with accessing the selfie_multiclass_256x256.tflite model file in the G:\AI\ComfyUI_M\ComfyUI\models\mediapipe directory. The system reports [Errno 13] Permission denied. Let’s step through this systematically:

---

Possible Causes & Solutions

1. File Permissions Issue:

   • The script is not allowed to read the file because of insufficient permissions.
   • Solution: Ensure the user executing the script has read permissions to the file and its parent directories.

   Steps:

   1. Right-click selfie_multiclass_256x256.tflite → Properties → Security.
   2. Ensure the user account or "Everyone" group has Read permissions.
   3. If not, click Edit → Add the relevant user → Check Read permission → Apply.

---

2. Windows Blocking the File:

   • Windows sometimes blocks downloaded files for security reasons.
   • Solution: Unblock the file.

   Steps:

   1. Right-click the selfie_multiclass_256x256.tflite file → Properties.
   2. Check if there's an "Unblock" button at the bottom → Click Unblock.

---

3. Antivirus or Security Software Interference:
   • Some antivirus programs block access to certain files by unfamiliar programs.
   • Solution: Temporarily disable the antivirus or create an exception for the ComfyUI directory.

---

4. Locked File:
   • The file could be in use by another process.
   • Solution: Restart your machine to release any potential file locks.

---

5. Remapping the Directory (Workaround):

   • Idea: If permission issues persist, you can copy the mediapipe folder to another location with full access and redirect the path programmatically using symbolic links or environment variables.

   Steps:

   1. Create a new folder: C:\Models\mediapipe.
   2. Copy selfie_multiclass_256x256.tflite into C:\Models\mediapipe.

   3. Create a symbolic link (Windows) to point to this new location:

      mklink /D "G:\AI\ComfyUI_M\ComfyUI\models\mediapipe" "C:\Models\mediapipe"

---

6. Execute with Elevated Privileges:

   • Solution: Run Python with administrator privileges.

   Steps:

   1. Open Command Prompt as Administrator.
   2. Navigate to your ComfyUI directory.

   3. Run the script:

      python main.py

---

Next Steps

• If none of these solutions work, there may be an issue with file corruption or disk access settings. Let me know which steps you’ve tried!

[neom315]

tried all of the above and all other nodes works without issues, but the problem is not resolve file can't be corrupted after downloading it 10 times and disk access is not giving issues on any other application.

about the 6th point and the python main.py return raise AssertionError("Torch not compiled with CUDA enabled") AssertionError: Torch not compiled with CUDA enabled

restarted comfyui, but nothing seems working

[neom315]

ok seems I have fixed thanks to trial and error

first take the selfie_multiclass_256x256.tflite and put it in the mediapipe folder, once you do that you will get an error about vitmatte missing, download the config.json preprocessor_config.json model.safetensors selfie_segmenter.tflite and put it in the vitmatte folder ( you have to create the folder manually if you don't have it) this is what made it work for me, you can get these models at

https://huggingface.co/hustvl/vitmatte-small-composition-1k/tree/main

[Awaken-ai]

vitmatte folder

sorry, tell me please the location of these folder

[bananaman1983]

ok seems I have fixed thanks to trial and error

first take the selfie_multiclass_256x256.tflite and put it in the mediapipe folder, once you do that you will get an error about vitmatte missing, download the config.json preprocessor_config.json model.safetensors selfie_segmenter.tflite and put it in the vitmatte folder ( you have to create the folder manually if you don't have it) this is what made it work for me, you can get these models at

https://huggingface.co/hustvl/vitmatte-small-composition-1k/tree/main

hey that one worked. TY mate

[frozenPG]

vitmatte folder

sorry, tell me please the location of these folder

you have to create a new directory in your models dir - "vitmatte" which contains files config.json, preprocessor_config.json, model.safetensors from link https://huggingface.co/hustvl/vitmatte-small-composition-1k/tree/main then in models/mediapipe there is a subdirectory "selfie_multiclass_256x256.tflite" which contains the file with the same name as the subdirectory "selfie_multiclass_256x256.tflite" - that is wrong. You need to put the file "selfie_multiclass_256x256.tflite" directly in models/mediapipe. Then it works - at least for me :)

[Jams63]

The error indicates a permission issue with accessing the selfie_multiclass_256x256.tflite model file in the G:\AI\ComfyUI_M\ComfyUI\models\mediapipe directory. The system reports [Errno 13] Permission denied. Let’s step through this systematically:

Possible Causes & Solutions

1. File Permissions Issue:

   • The script is not allowed to read the file because of insufficient permissions.
   • Solution: Ensure the user executing the script has read permissions to the file and its parent directories.

   Steps:

   1. Right-click selfie_multiclass_256x256.tflite → Properties → Security.
   2. Ensure the user account or "Everyone" group has Read permissions.
   3. If not, click Edit → Add the relevant user → Check Read permission → Apply.

2. Windows Blocking the File:

   • Windows sometimes blocks downloaded files for security reasons.
   • Solution: Unblock the file.

   Steps:

   1. Right-click the selfie_multiclass_256x256.tflite file → Properties.
   2. Check if there's an "Unblock" button at the bottom → Click Unblock.

3. Antivirus or Security Software Interference:

   • Some antivirus programs block access to certain files by unfamiliar programs.
   • Solution: Temporarily disable the antivirus or create an exception for the ComfyUI directory.

4. Locked File:

   • The file could be in use by another process.
   • Solution: Restart your machine to release any potential file locks.

5. Remapping the Directory (Workaround):

   • Idea: If permission issues persist, you can copy the mediapipe folder to another location with full access and redirect the path programmatically using symbolic links or environment variables.

   Steps:

   1. Create a new folder: C:\Models\mediapipe.
   2. Copy selfie_multiclass_256x256.tflite into C:\Models\mediapipe.
   3. Create a symbolic link (Windows) to point to this new location:

      mklink /D "G:\AI\ComfyUI_M\ComfyUI\models\mediapipe" "C:\Models\mediapipe"

6. Execute with Elevated Privileges:

   • Solution: Run Python with administrator privileges.

   Steps:

   1. Open Command Prompt as Administrator.
   2. Navigate to your ComfyUI directory.
   3. Run the script:

      python main.py

Next Steps

• If none of these solutions work, there may be an issue with file corruption or disk access settings. Let me know which steps you’ve tried!

Unblocking and remapping method worked for me , thanks