Cellprofiler 4.0.7 and CellPose 3.0.7 - RunCellpose is out of date

[kmshort]

Firstly, my versioning information:

CellProfiler version:   4.2.7
cellpose version:       3.0.7
platform:               win32
python version:         3.8.19
torch version:          2.4.0
conda version:          24.7.1

I am using the Cuda SDK v12.4, and torch version 2.4.0

I am using a conda environment with both CellProfiler and Cellpose running in it. I have installed the GUI version of CellPose, and it runs well with CUDA/GPU processing enabled.

However, there are issues with the RunCellpose plugin in CellProfiler. runcellpose.py was pulled via git clone today, 8th August 2024.

Issues:

1. In the CellProfiler project pipeline, the RunCellpose items are crossed ❌ with the "Value not in decimal format" notification. With the workaround I have, that issue doesn't affect the processing.

2. The models in the Cellpose 2.x are now mostly removed, but are referenced by the current runcellpose.py.

3. net_avg used by the plugin is no longer used by CellPose 3.0.7, so the plugin throws an error and exits.

(cellpose) C:\Users\user\anaconda3\envs\cellpose\Scripts\CellProfiler_Plugins>cellprofiler
Progress Counter({'Unprocessed': 1})
Starting workers on address tcp://127.0.0.1:62355
Worker 0: C:\Users\usert\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\worker\__init__.py:25: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
Worker 0:   import pkg_resources
Worker 0: Starting job
Progress Counter({'InProcess': 1})
Worker 0: Doing job: 1
Worker 0: Running module Images 1
Worker 0: Thu Aug  8 11:31:18 2024: Image # 1, module Images # 1: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module Metadata 2
Worker 0: Thu Aug  8 11:31:18 2024: Image # 1, module Metadata # 2: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module NamesAndTypes 3
Worker 0: Thu Aug  8 11:31:18 2024: Image # 1, module NamesAndTypes # 3: CPU_time = 0.25 secs, Wall_time = 0.50 secs
Worker 0: Running module Groups 4
Worker 0: Thu Aug  8 11:31:19 2024: Image # 1, module Groups # 4: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module RunCellpose 5
Worker 0: >>>> loading model c:\users\user\Desktop\CellPose Models\StellarisDapi_20240807_131156
Worker 0: >>>> using CPU
Worker 0: C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellpose\resnet_torch.py:298: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
Worker 0:   state_dict = torch.load(filename, map_location=torch.device("cpu"))
Worker 0: >>>> model diam_mean =  30.000 (ROIs rescaled to this size during training)
Worker 0: >>>> model diam_labels =  31.044 (mean diameter of training ROIs)
Worker 0: Unable to create masks. Check your module settings. eval() got an unexpected keyword argument 'net_avg'
Worker 0: run_image_set_exception, get_always_continue False
Worker 0: Error detected during run of module RunCellpose#5
Worker 0: Traceback (most recent call last):
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\pipeline\_pipeline.py", line 1208, in run_image_set
Worker 0:     self.run_module(module, workspace)
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\pipeline\_pipeline.py", line 1349, in run_module
Worker 0:     module.run(workspace)
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\Scripts\CellProfiler_Plugins\CellProfiler-plugins\active_plugins\runcellpose.py", line 638, in run
Worker 0:     y.segmented = y_data
Worker 0: UnboundLocalError: local variable 'y_data' referenced before assignment
Progress Counter({'FinishedWaitingMeasurements': 1})
Progress Counter({'Done': 1})
Finished!
Worker 0: Cancelling worker

If I comment out net_avg lines 536, 551, and 610. The plugin works, but with CPU only (see next).

4. The Test GPU button does not work. The GPU detection fails (the reference to core.use_gpu on line 717 in the script fails, because there's no reference to core:

File "C:\Users\user\anaconda3\envs\cellpose\Scripts\CellProfiler_Plugins\CellProfiler-plugins\active_plugins\runcellpose.py", line 722, in do_check_gpu
    GPU_works = core.use_gpu(use_torch=torch_installed)
NameError: name 'core' is not defined

5. When using the GPU anyway, the process fails even though that there is a line that says: ****Worker 0: TORCH CUDA version installed and working. ****:

Progress Counter({'Unprocessed': 1})
Starting workers on address tcp://127.0.0.1:64668
Worker 0: C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\worker\__init__.py:25: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
Worker 0:   import pkg_resources
Worker 0: Starting job
Progress Counter({'InProcess': 1})
Worker 0: Doing job: 1
Worker 0: Running module Images 1
Worker 0: Thu Aug  8 11:44:16 2024: Image # 1, module Images # 1: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module Metadata 2
Worker 0: Thu Aug  8 11:44:16 2024: Image # 1, module Metadata # 2: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module NamesAndTypes 3
Worker 0: Thu Aug  8 11:44:16 2024: Image # 1, module NamesAndTypes # 3: CPU_time = 0.38 secs, Wall_time = 0.49 secs
Worker 0: Running module Groups 4
Worker 0: Thu Aug  8 11:44:16 2024: Image # 1, module Groups # 4: CPU_time = 0.00 secs, Wall_time = 0.00 secs
Worker 0: Running module RunCellpose 5
Worker 0: >>>> loading model c:\users\userDesktop\CellPose Models\StellarisDapi_20240807_131156
Worker 0: ** TORCH CUDA version installed and working. **
Worker 0: >>>> using GPU
Worker 0: C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellpose\resnet_torch.py:294: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
Worker 0:   state_dict = torch.load(filename, map_location=device)
Worker 0: >>>> model diam_mean =  30.000 (ROIs rescaled to this size during training)
Worker 0: >>>> model diam_labels =  31.044 (mean diameter of training ROIs)
Worker 0: run_image_set_exception, get_always_continue False
Worker 0: Error detected during run of module RunCellpose#5
Worker 0: Traceback (most recent call last):
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\pipeline\_pipeline.py", line 1208, in run_image_set
Worker 0:     self.run_module(module, workspace)
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\lib\site-packages\cellprofiler_core\pipeline\_pipeline.py", line 1349, in run_module
Worker 0:     module.run(workspace)
Worker 0:   File "C:\Users\user\anaconda3\envs\cellpose\Scripts\CellProfiler_Plugins\CellProfiler-plugins\active_plugins\runcellpose.py", line 526, in run
Worker 0:     if self.use_gpu.value and model.torch:
Worker 0: AttributeError: 'CellposeModel' object has no attribute 'torch'
Worker 0: Cancelling worker
Worker 0: Aborting job after cancellation
Cancelled!

So in summary, there are a few major issues with runcellpose.py and CellPose 3.0.7. My current edits to remove net_avg, make it work, but only via CPU. I am using custom trained models generated in CellPose 3.0.7.

I hope these issues can be resolved. It would be good to get GPU processing working. I'll continue to look at the python script to see what I can do.

[kmshort]

And the available switches for CellPose 3.0.7:

usage: cellpose [-h] [--version] [--verbose] [--Zstack] [--use_gpu] [--gpu_device GPU_DEVICE] [--check_mkl]
                [--dir DIR] [--image_path IMAGE_PATH] [--look_one_level_down] [--img_filter IMG_FILTER]
                [--channel_axis CHANNEL_AXIS] [--z_axis Z_AXIS] [--chan CHAN] [--chan2 CHAN2] [--invert]
                [--all_channels] [--pretrained_model PRETRAINED_MODEL] [--restore_type RESTORE_TYPE] [--chan2_restore]
                [--add_model ADD_MODEL] [--no_resample] [--no_interp] [--no_norm] [--do_3D] [--diameter DIAMETER]
                [--stitch_threshold STITCH_THRESHOLD] [--min_size MIN_SIZE] [--flow_threshold FLOW_THRESHOLD]
                [--cellprob_threshold CELLPROB_THRESHOLD] [--niter NITER] [--anisotropy ANISOTROPY]
                [--exclude_on_edges] [--augment] [--save_png] [--save_tif] [--no_npy] [--savedir SAVEDIR]
                [--dir_above] [--in_folders] [--save_flows] [--save_outlines] [--save_rois] [--save_txt] [--save_mpl]
                [--train] [--train_size] [--test_dir TEST_DIR] [--mask_filter MASK_FILTER] [--diam_mean DIAM_MEAN]
                [--learning_rate LEARNING_RATE] [--weight_decay WEIGHT_DECAY] [--n_epochs N_EPOCHS]
                [--batch_size BATCH_SIZE] [--min_train_masks MIN_TRAIN_MASKS] [--SGD SGD] [--save_every SAVE_EVERY]
                [--model_name_out MODEL_NAME_OUT]

Cellpose Command Line Parameters

optional arguments:
  -h, --help            show this help message and exit
  --version             show cellpose version info
  --verbose             show information about running and settings and save to log
  --Zstack              run GUI in 3D mode

Hardware Arguments:
  --use_gpu             use gpu if torch with cuda installed
  --gpu_device GPU_DEVICE
                        which gpu device to use, use an integer for torch, or mps for M1
  --check_mkl           check if mkl working

Input Image Arguments:
  --dir DIR             folder containing data to run or train on.
  --image_path IMAGE_PATH
                        if given and --dir not given, run on single image instead of folder (cannot train with this
                        option)
  --look_one_level_down
                        run processing on all subdirectories of current folder
  --img_filter IMG_FILTER
                        end string for images to run on
  --channel_axis CHANNEL_AXIS
                        axis of image which corresponds to image channels
  --z_axis Z_AXIS       axis of image which corresponds to Z dimension
  --chan CHAN           channel to segment; 0: GRAY, 1: RED, 2: GREEN, 3: BLUE. Default: 0
  --chan2 CHAN2         nuclear channel (if cyto, optional); 0: NONE, 1: RED, 2: GREEN, 3: BLUE. Default: 0
  --invert              invert grayscale channel
  --all_channels        use all channels in image if using own model and images with special channels

Model Arguments:
  --pretrained_model PRETRAINED_MODEL
                        model to use for running or starting training
  --restore_type RESTORE_TYPE
                        model to use for image restoration
  --chan2_restore       use nuclei restore model for second channel
  --add_model ADD_MODEL
                        model path to copy model to hidden .cellpose folder for using in GUI/CLI

Algorithm Arguments:
  --no_resample         disable dynamics on full image (makes algorithm faster for images with large diameters)
  --no_interp           do not interpolate when running dynamics (was default)
  --no_norm             do not normalize images (normalize=False)
  --do_3D               process images as 3D stacks of images (nplanes x nchan x Ly x Lx
  --diameter DIAMETER   cell diameter, if 0 will use the diameter of the training labels used in the model, or with
                        built-in model will estimate diameter for each image
  --stitch_threshold STITCH_THRESHOLD
                        compute masks in 2D then stitch together masks with IoU>0.9 across planes
  --min_size MIN_SIZE   minimum number of pixels per mask, can turn off with -1
  --flow_threshold FLOW_THRESHOLD
                        flow error threshold, 0 turns off this optional QC step. Default: 0.4
  --cellprob_threshold CELLPROB_THRESHOLD
                        cellprob threshold, default is 0, decrease to find more and larger masks
  --niter NITER         niter, number of iterations for dynamics for mask creation, default of 0 means it is
                        proportional to diameter, set to a larger number like 2000 for very long ROIs
  --anisotropy ANISOTROPY
                        anisotropy of volume in 3D
  --exclude_on_edges    discard masks which touch edges of image
  --augment             tiles image with overlapping tiles and flips overlapped regions to augment

Output Arguments:
  --save_png            save masks as png and outlines as text file for ImageJ
  --save_tif            save masks as tif and outlines as text file for ImageJ
  --no_npy              suppress saving of npy
  --savedir SAVEDIR     folder to which segmentation results will be saved (defaults to input image directory)
  --dir_above           save output folders adjacent to image folder instead of inside it (off by default)
  --in_folders          flag to save output in folders (off by default)
  --save_flows          whether or not to save RGB images of flows when masks are saved (disabled by default)
  --save_outlines       whether or not to save RGB outline images when masks are saved (disabled by default)
  --save_rois           whether or not to save ImageJ compatible ROI archive (disabled by default)
  --save_txt            flag to enable txt outlines for ImageJ (disabled by default)
  --save_mpl            save a figure of image/mask/flows using matplotlib (disabled by default). This is slow,
                        especially with large images.

Training Arguments:
  --train               train network using images in dir
  --train_size          train size network at end of training
  --test_dir TEST_DIR   folder containing test data (optional)
  --mask_filter MASK_FILTER
                        end string for masks to run on. use '_seg.npy' for manual annotations from the GUI. Default:
                        _masks
  --diam_mean DIAM_MEAN
                        mean diameter to resize cells to during training -- if starting from pretrained models it
                        cannot be changed from 30.0
  --learning_rate LEARNING_RATE
                        learning rate. Default: 0.2
  --weight_decay WEIGHT_DECAY
                        weight decay. Default: 1e-05
  --n_epochs N_EPOCHS   number of epochs. Default: 500
  --batch_size BATCH_SIZE
                        batch size. Default: 8
  --min_train_masks MIN_TRAIN_MASKS
                        minimum number of masks a training image must have to be used. Default: 5
  --SGD SGD             use SGD
  --save_every SAVE_EVERY
                        number of epochs to skip between saves. Default: 100
  --model_name_out MODEL_NAME_OUT
                        Name of model to save as, defaults to name describing model architecture. Model is saved in
                        the folder specified by --dir in models subfolder.