[Bug]: Preprocess images in the "Train" tab crashes when "Auto Focal Point Crop" is selected

[green-mint]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits

What happened?

The preprocessor crashes when detecting faces using the OpenCV library.

Steps to reproduce the problem

1. Selecting only the "Auto Focal Point Crop" option in the train tab.
2. Click Preprocess

What should have happened?

The images should have been preprocessed without errors

Version or Commit where the problem happens

1.4.0

What Python version are you running on ?

Python 3.10.x

What platforms do you use to access the UI ?

Windows

What device are you running WebUI on?

Nvidia GPUs (GTX 16 below)

Cross attention optimization

Automatic

What browsers do you use to access the UI ?

Google Chrome

Command Line Arguments

No modification to the bat file or the cli arguments

List of extensions

I have installed sd_dreambooth_extension (manual installation, cloned the repo, installed the packages after activating the venv) but the issue persisted even after disabling the extension.

Console logs

Preprocessing [Image 0/1]:   0%|                                                                 | 0/1 [00:00<?, ?it/s]
*** Error completing request
*** Arguments: ('task(1d2hiirzdek2l1f)', 'C:\\Users\\EternalBhutto\\saad\\imgs', 'C:\\Users\\EternalBhutto\\saad\\imgs-new', 512, 512, 'ignore', False, False, False, False, False, 0.5, 0.2, True, 0.9, 0.15, 0.5, False, False, 384, 768, 4096, 409600, 'Maximize area', 0.1) {}
    Traceback (most recent call last):
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\call_queue.py", line 55, in f
        res = list(func(*args, **kwargs))
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\call_queue.py", line 35, in f
        res = func(*args, **kwargs)
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\ui.py", line 19, in preprocess
        modules.textual_inversion.preprocess.preprocess(*args)
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\preprocess.py", line 18, in preprocess
        preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_keep_original_size, process_flip, process_split, process_caption, process_caption_deepbooru, split_threshold, overlap_ratio, process_focal_crop, process_focal_crop_face_weight, process_focal_crop_entropy_weight, process_focal_crop_edges_weight, process_focal_crop_debug, process_multicrop, process_multicrop_mindim, process_multicrop_maxdim, process_multicrop_minarea, process_multicrop_maxarea, process_multicrop_objective, process_multicrop_threshold)
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\preprocess.py", line 212, in preprocess_work
        for focal in autocrop.crop_image(img, autocrop_settings):
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\autocrop.py", line 32, in crop_image
        focus = focal_point(im_debug, settings)
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\autocrop.py", line 75, in focal_point
        face_points = image_face_points(im, settings) if settings.face_points_weight > 0 else []
      File "C:\Users\EternalBhutto\saad\stable-diffusion-webui\modules\textual_inversion\autocrop.py", line 150, in image_face_points
        faces = detector.detect(np.array(im))
    cv2.error: OpenCV(4.8.0) D:\a\opencv-python\opencv-python\opencv\modules\dnn\src\net_impl.cpp:279: error: (-204:Requested object was not found) Layer with requested id=-1 not found in function 'cv::dnn::dnn4_v20230620::Net::Impl::getLayerData'


---

Additional information

No response

[Cosmos01]

run C:\Users\EternalBhutto\saad\stable-diffusion-webui\venv\Scripts\activate.bat
(venv) > pip install opencv-python==4.7.0.72

[202030481266]

If you using linux (such as ubuntu 18.04), you should do the following steps:

1. source stable-diffusion-webui/venv/bin/activate
2. pip install opencv-python==4.7.0.72

and restart webui.sh.

You should know that the webui runs in stable-diffusion-webui/venv virtual environment when you start it by running webui.sh.

[green-mint]

run C:\Users\EternalBhutto\saad\stable-diffusion-webui\venv\Scripts\activate.bat
(venv) > pip install opencv-python==4.7.0.72

Thankyou that worked 🎉

[TeKett]

pip install opencv-python==4.7.0.72

Same problem, tried this, and it says
Requirement already satisfied: opencv-python==4.7.0.72 in c:\users\user\stable-diffusion-webui\venv\lib\site-packages (4.7.0.72)
Requirement already satisfied: numpy>=1.17.0 in c:\users\user\stable-diffusion-webui\venv\lib\site-packages (from opencv-python==4.7.0.72) (1.23.5)

Since it last worked, i do have upgraded a1111 from 1.3.2 to 1.4, and now to 1.5 as it was just released 3 hours ago.

I also tried to delete the cv2 and opencv-4.7.0.72 (which was just a info folder, but would not let me install otherwise) folder, and then did the install, and still same problem, both version 4.7.0.72 and 4.8.0.74.

edit: Is it worth mentioning that the opencv version the error claim i have is wrong and that the file it says has the error in it dont exist? Like where is this even supposed to be "D:\a\opencv-python\opencv-python\opencv\modules\dnn\src\net_impl.cpp", its certainly not on my drive at least.

[steven850]

I am getting the same error when trying to preprocess with focal crop, im using the 1.4.0 release package.
I also do not have a D drive, yet its looking for the file in the same location as you mentioned.

Traceback (most recent call last): File "Z:\SD\webui\modules\call_queue.py", line 55, in f res = list(func(*args, **kwargs)) File "Z:\SD\webui\modules\call_queue.py", line 35, in f res = func(*args, **kwargs) File "Z:\SD\webui\modules\textual_inversion\ui.py", line 19, in preprocess modules.textual_inversion.preprocess.preprocess(*args) File "Z:\SD\webui\modules\textual_inversion\preprocess.py", line 18, in preprocess preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_keep_original_size, process_flip, process_split, process_caption, process_caption_deepbooru, split_threshold, overlap_ratio, process_focal_crop, process_focal_crop_face_weight, process_focal_crop_entropy_weight, process_focal_crop_edges_weight, process_focal_crop_debug, process_multicrop, process_multicrop_mindim, process_multicrop_maxdim, process_multicrop_minarea, process_multicrop_maxarea, process_multicrop_objective, process_multicrop_threshold) File "Z:\SD\webui\modules\textual_inversion\preprocess.py", line 212, in preprocess_work for focal in autocrop.crop_image(img, autocrop_settings): File "Z:\SD\webui\modules\textual_inversion\autocrop.py", line 32, in crop_image focus = focal_point(im_debug, settings) File "Z:\SD\webui\modules\textual_inversion\autocrop.py", line 75, in focal_point face_points = image_face_points(im, settings) if settings.face_points_weight > 0 else [] File "Z:\SD\webui\modules\textual_inversion\autocrop.py", line 150, in image_face_points faces = detector.detect(np.array(im)) cv2.error: OpenCV(4.8.0) D:\a\opencv-python\opencv-python\opencv\modules\dnn\src\net_impl.cpp:279: error: (-204:Requested object was not found) Layer with requested id=-1 not found in function 'cv::dnn::dnn4_v20230620::Net::Impl::getLayerData'

[ruhtel]

Try removing command line arguments if you're getting this error even after the new opencv-python version.

I had --xformers and --listen, removing these fixed the issue.

EDIT: I retried with --xformers and it's working, either --listen is the issue or it was a crazy coincidence.

[tlegower]

If you are running the extension ControlNet, it will uninstall OpenCV-Python 4.7 and install 4.8. I fixed this issue by:

Turn off ControlNet extension in Webui
Uninstall OpenCV-Python 4.8
Install OpenCV-Python 4.7.0.72
Then I ran launch.py and training with face focal point is running.

[w-e-w]

Fix PR #14121