MouseLand / cellpose

a generalist algorithm for cellular segmentation with human-in-the-loop capabilities
https://www.cellpose.org/
BSD 3-Clause "New" or "Revised" License
1.24k stars 359 forks source link

[BUG] Cellpose 3.0.7 does not remember that turning off auto-adjust saturation during image training #920

Closed leon11345e closed 1 month ago

leon11345e commented 2 months ago

Describe the bug Cellpose 3.0.7 does not remember that turning off auto-adjust saturation during image training.

To Reproduce Steps to reproduce the behavior:

  1. deactivate auto-adjust saturation
  2. perform training on image
  3. newly loaded image has autoadjusted saturation on again

also happens without training if any type of model is used to segment image

Logs:

(cellpose) C:\Users\leone>Cellpose 2024-04-18 15:25:55,577 [INFO] WRITING LOG OUTPUT TO C:\Users\leone.cellpose\run.log 2024-04-18 15:25:55,577 [INFO] cellpose version: 3.0.7 platform: win32 python version: 3.8.19 torch version: 2.2.2+cu118 2024-04-18 15:25:55,792 [INFO] TORCH CUDA version installed and working. GUI_INFO: loading image: C:/Users/leone/Documents/SingleCell Local/WT mEOS Comparison/M1 Choosen/Training 18.4.24 Flowcell WT mEOS + WT vom 20.10.23/mEOS6-94.tif GUI_INFO: selected model new Training flowcell+wt old, loading now 2024-04-18 15:28:03,801 [INFO] >> new Training flowcell+wt old << model set to be used 2024-04-18 15:28:03,801 [INFO] TORCH CUDA version installed and working. 2024-04-18 15:28:03,801 [INFO] >>>> using GPU 2024-04-18 15:28:03,886 [INFO] >>>> model diam_mean = 30.000 (ROIs rescaled to this size during training) 2024-04-18 15:28:03,886 [INFO] >>>> model diam_labels = 21.940 (mean diameter of training ROIs) GUI_INFO: diameter set to 21.94 (but can be changed) 2024-04-18 15:28:05,757 [INFO] >>>> loading model C:\Users\leone.cellpose\models\new Training flowcell+wt old 2024-04-18 15:28:05,776 [INFO] TORCH CUDA version installed and working. 2024-04-18 15:28:05,776 [INFO] >>>> using GPU 2024-04-18 15:28:05,857 [INFO] >>>> model diam_mean = 30.000 (ROIs rescaled to this size during training) 2024-04-18 15:28:05,857 [INFO] >>>> model diam_labels = 21.940 (mean diameter of training ROIs) {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} 2024-04-18 15:28:07,621 [INFO] 7 cells found with model in 1.863 sec GUI_INFO: 7 masks found GUI_INFO: creating cellcolors and drawing masks {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} [0, 255.0] 2024-04-18 15:28:16,551 [INFO] >>>> loading model C:\Users\leone.cellpose\models\new Training flowcell+wt old 2024-04-18 15:28:16,551 [INFO] TORCH CUDA version installed and working. 2024-04-18 15:28:16,551 [INFO] >>>> using GPU 2024-04-18 15:28:16,651 [INFO] >>>> model diam_mean = 30.000 (ROIs rescaled to this size during training) 2024-04-18 15:28:16,651 [INFO] >>>> model diam_labels = 21.940 (mean diameter of training ROIs) {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} 2024-04-18 15:28:17,065 [INFO] 7 cells found with model in 0.513 sec GUI_INFO: 7 masks found GUI_INFO: creating cellcolors and drawing masks {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} [0, 255.0] 2024-04-18 15:28:22,510 [INFO] TORCH CUDA version installed and working. 2024-04-18 15:28:22,510 [INFO] >>>> using GPU 2024-04-18 15:28:22,517 [INFO] >> cyto3 << model set to be used 2024-04-18 15:28:22,595 [INFO] >>>> model diam_mean = 30.000 (ROIs rescaled to this size during training) {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} 2024-04-18 15:28:22,595 [INFO] channels set to [0, 0] 2024-04-18 15:28:22,611 [INFO] ~ FINDING MASKS ~ 2024-04-18 15:28:23,012 [INFO] >>>> TOTAL TIME 0.42 sec 2024-04-18 15:28:23,018 [INFO] 7 cells found with model in 0.508 sec GUI_INFO: 7 masks found GUI_INFO: creating cellcolors and drawing masks {'lowhigh': None, 'percentile': [1.0, 99.0], 'normalize': True, 'norm3D': False, 'sharpen_radius': 0, 'smooth_radius': 0, 'tile_norm_blocksize': 0, 'tile_norm_smooth3D': 1, 'invert': False} [0, 255.0]

carsen-stringer commented 1 month ago

thanks for reporting this, it should be fixed now: pip install git+https://github.com/mouseland/cellpose.git