downloads:cryolo_1

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
downloads:cryolo_1 [2019/07/11 11:39]
twagner [crYOLO boxmanager]
downloads:cryolo_1 [2019/10/30 14:16] (current)
twagner [Installation]
Line 24: Line 24:
 ====== Download ====== ====== Download ======
  
- +<note important>​
- +
----- +
 Before downloading or using this product, make sure you **understand and accept the [[:​cryolo_license|terms of the license]]**. ​ Before downloading or using this product, make sure you **understand and accept the [[:​cryolo_license|terms of the license]]**. ​
 +</​note>​
  
----- 
  
  
 ====crYOLO==== ====crYOLO====
-Version: 1.4.0 
  
-Uploaded:  ​11July 2019+Version: 1.5.4 
 + 
 +Uploaded:  ​30October ​2019
  
-[[ftp://ftp.gwdg.de/pub/misc/sphire/​crYOLO_V1_4_0/​cryolo-1.4.0.tar.gz|DOWNLOAD]]+[[https://pypi.org/project/cryolo/#files|DOWNLOAD]]
  
 Please see [[downloads:​cryolo_1#​installation|install instructions]] how to get it running on the CPU. Please see [[downloads:​cryolo_1#​installation|install instructions]] how to get it running on the CPU.
Line 44: Line 42:
 ====crYOLO boxmanager==== ====crYOLO boxmanager====
  
-Version: 1.2.3+Version: 1.2.7
  
-Uploaded: ​11July 2019+Uploaded: ​26September ​2019
  
-[[ftp://ftp.gwdg.de/pub/misc/sphire/​crYOLO_BM_V1_2_3/​cryoloBM-1.2.3.tar.gz|DOWNLOAD]]+[[https://pypi.org/project/cryoloBM/#files|DOWNLOAD]]
  
 [{{ :​downloads:​cryolophosaurusdb.jpg?​150|**crYOLO Phosauraus**Net'​s eponym}}] [{{ :​downloads:​cryolophosaurusdb.jpg?​150|**crYOLO Phosauraus**Net'​s eponym}}]
Line 54: Line 52:
 ==== General PhosaurusNet models ==== ==== General PhosaurusNet models ====
  
 +[[pipeline:​window:​cryolo:​picking_general|Please see the tutorial how to use the general models]]
  
 === For cryo images (low-pass filtered) === === For cryo images (low-pass filtered) ===
-Number of datasets: ​38 real, 10 simulated, 10 particle free datasets on various grids with contamination+Number of datasets: ​41 real, 10 simulated, 10 particle free datasets on various grids with contamination
  
-Uploaded: ​17May 2019+Uploaded: ​15Oktober ​2019
  
-[[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_20190516.h5+[[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_201910.h5
 |DOWNLOAD]] ​ |DOWNLOAD]] ​
- 
-[[pipeline:​window:​cryolo#​picking_particles_-_without_training_using_a_general_model |Valid configuration file]] 
  
 === For cryo images (neural network denoised with JANNI) === === For cryo images (neural network denoised with JANNI) ===
-Number of datasets: ​38 real, 10 simulated, 10 particle free datasets on various grids with contamination+Number of datasets: ​41 real, 10 simulated, 10 particle free datasets on various grids with contamination
  
-Uploaded: ​11July 2019+Uploaded: ​15Oktober ​2019
  
-[[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_201907_nn_denoise.h5+[[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_201910_nn_denoise.h5
 |DOWNLOAD]] ​ |DOWNLOAD]] ​
- 
-[[pipeline:​window:​cryolo#​cryoem_images |Valid configuration file]] 
  
 <​note>​ <​note>​
Line 88: Line 83:
 [[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_negstain_20190226.h5 [[ftp://​ftp.gwdg.de/​pub/​misc/​sphire/​crYOLO-GENERAL-MODELS/​gmodel_phosnet_negstain_20190226.h5
 |DOWNLOAD]] ​ |DOWNLOAD]] ​
- 
-[[pipeline:​window:​cryolo#​negative_stain_images |Valid configuration file]] 
 ====ARCHIVE==== ====ARCHIVE====
  
 Previous versions of crYOLO, the boxmanager and the general models can be found here: [[http://​sphire.mpg.de/​wiki/​doku.php?​id=cryolo_archive#​cryolo|Archive]]. Previous versions of crYOLO, the boxmanager and the general models can be found here: [[http://​sphire.mpg.de/​wiki/​doku.php?​id=cryolo_archive#​cryolo|Archive]].
- 
-====== Known issues ===== 
-  * Issue 0: Training on multiple GPUs sometimes lead to worse performance (higher loss). We currently recommend to train on single gpus. 
-  * Issue 17: On the fly filtering (%%--%%otf) is slower than using it not, as the filtering is not parallelized in this case.  
-<hidden Closed issues> 
-  * <​del> ​ Issue 1: crYOLO sometimes not exit properly after training finished. Has to be killed manually.</​del>​ 
-  * <​del>​Issue 2: If you use automatic filtering with .tif files, you get an error like "​OSError:​ cannot identify image file '​filtered_folder/​another_folder/​my_image.tif'"​. It will be fixed in the next release.</​del>​ 
-  * <​del>​Issue 3: (Boxmanager) The visualization only shows the first filament when loading eman1 helical box files (start end coordinates). Will be fixed in the next release.</​del>​ 
-  * <​del>​Issue 4: The filament mode will crash if crYOLO cannot identify a single particle in the image. Will be fixed in 1.2.2</​del>​ 
-  * <​del>​Issue 5: If movies were aligned with cisTEM and picked with crYOLO, the box position are vertically flipped. Will be fixed in 1.2.2</​del>​ 
-  * <​del>​Issue 6: crYOLO does overwrite the environmental variable "​CUDA_VISIBLE_DEVICES"​ with 0 if no gpu is specified by the -g parameter. This leads to the behavior that crYOLO ignores previous settings in CUDA_VISIBLE_DEVICES. Will be fixed in 1.2.2</​del>​ 
-  * <​del>​Issue 7: On K3 images crYOLO seems to add a offset toward the longer axis of the input image.</​del>​ 
-  * <​del>​Issue 8: There is a logical error in filament tracing, which sometimes connects two parallel filaments.</​del>​ 
-  * <​del>​Issue 9: Some people report an error when running cryolo prediction/​training:​ "​ImportError:​ numpy.core.multiarray failed to import"​. It will be fixed in 1.2.3.</​del>​ 
-  * <​del>​Issue 10: On machines with many cores (e.g 64) an error during filtering might pop up: "​[ERROR:​0] 53: Can't spawn new thread"</​del>​ 
-  * <​del>​Issue 11: If the -g parameter is not provided, crYOLO will use the memory of all GPUs. Will be fixed in 1.2.3.</​del>​ 
-  * <​del>​Issue 12: The LineEnhancer depdenceny of crYOLO is still dependent from opencv. Workaround: In the crYOLO environment:​ conda install opencv</​del>​ 
-  * <​del>​Issue 13: After picking it can happen that some of the boxes are not fully immersed in the image. Will be fixed in 1.2.4.</​del>​ 
-  * <​del>​Issue 14: Parallelization in filament mode is broken. Will be fixed in 1.2.4.</​del>​ 
-  * <​del>​Issue 15: If the %%--%%gpu_fraction is used, crYOLO always uses GPU 0. Will be fixed in 1.3.1.</​del> ​ 
-  * <​del>​Issue 16: %%--%%gpu_fraction only works for prediction, not for training. Will be fixed in 1.3.2.</​del>​ 
-  * <​del>​Issue 18: Prediction is broken in 1.3.2. It removes all particles as it claim they are not fully immersed in the image.</​del>​ 
-  * <​del>​Issue 19: Filtering does not work if target image directory is absolute path.</​del>​ 
-  * <​del>​Issue 20: crYOLO 1.3.4 has a normalization bug. During training the images are normalized seperately, but during prediction is done batch wise. Workaround: Use -pbs 1 during prediction. It will be fixed in 1.3.5.</​del>​ 
-  * <​del>​Issue 21: The search range for filament tracing is too low for many datasets. To check if you are affected: Use your trained model and pick without the filament options. Check if your filaments a nicely picked (many consecutive boxes on a filament). In the next version, the search range will be increased and added as an optional parameter.</​del>​ 
-  * <​del>​Issue 22: If absolute paths are used in the field "​train_image"​ in your configuration file, filtering is skipped.</​del>​ 
-</​hidden>​ 
  
  
Line 145: Line 111:
 After that, create a new virtual environment:​ After that, create a new virtual environment:​
 <​code>​ <​code>​
-conda create -n cryolo -c anaconda python=3.6 pyqt=5 cudnn=7.1.2 cython+conda create -n cryolo -c anaconda python=3.6 pyqt=5 cudnn=7.1.2 ​numpy==1.14.5 ​cython ​wxPython==4.0.4 intel-openmp==2019.4
 </​code>​ </​code>​
  
Line 151: Line 117:
 <​code>​ <​code>​
 source activate cryolo source activate cryolo
-</​code>​ 
- 
-Install fast numpy from conda: 
-<​code>​ 
-conda install numpy==1.15.4 
 </​code>​ </​code>​
  
 In case you run **crYOLO on a GPU** run: In case you run **crYOLO on a GPU** run:
 <​code>​ <​code>​
-pip install cryolo-X.Y.Z.tar.gz[gpu] +pip install ​'cryolo[gpu]' ​
 </​code>​ </​code>​
  
 But if you want to run **crYOLO on a CPU** run: But if you want to run **crYOLO on a CPU** run:
 <​code>​ <​code>​
-pip install cryolo-X.Y.Z.tar.gz[cpu] +pip install ​'cryolo[cpu]' ​
 </​code>​ </​code>​
  
-Finally ​you install ​the cryolo ​boxmanager+<​note>​ 
-<code+During the installation of crYOLO ​you will see the following error message: 
-pip install ​cryoloBM-X.Y.Z.tar.gz +''​ERROR:​ imagecodecs-lite 2019.2.22 has requirement numpy>​=1.15.4,​ but you'll have numpy 1.14.5 which is incompatible.''​ 
-</code>+However, you can ignore it. It is actually also working with numpy==1.14.5 
 +</​note>​ 
 + 
 +<note tip> 
 +In case you want to integrate crYOLO in your module system, [[pipeline:​window:​cryolo:integration_into_module_system|here is a brief explanation]]. 
 +</note
 + 
 + 
 +<note important>​ 
 +**Known issues** 
 + 
 +  * Issue 0: Training on multiple GPUs sometimes lead to worse performance (higher loss). We currently recommend to train on single gpus. 
 +  * Issue 17: On the fly filtering (%%--%%otf) is slower than using it not, as the filtering is not parallelized in this case.  
 + 
 + 
 + 
 +<hidden Closed issues>​ 
 +  * <​del> ​ Issue 1: crYOLO sometimes not exit properly after training finished. Has to be killed manually.</​del>​ 
 +  * <​del>​Issue 2: If you use automatic filtering with .tif files, you get an error like "​OSError:​ cannot identify image file '​filtered_folder/​another_folder/​my_image.tif'"​. It will be fixed in the next release.</​del>​ 
 +  * <​del>​Issue 3: (Boxmanager) The visualization only shows the first filament when loading eman1 helical box files (start end coordinates). Will be fixed in the next release.</​del>​ 
 +  * <​del>​Issue 4: The filament mode will crash if crYOLO cannot identify a single particle in the image. Will be fixed in 1.2.2</​del>​ 
 +  * <​del>​Issue 5: If movies were aligned with cisTEM and picked with crYOLO, the box position are vertically flipped. Will be fixed in 1.2.2</​del>​ 
 +  * <​del>​Issue 6: crYOLO does overwrite the environmental variable "​CUDA_VISIBLE_DEVICES"​ with 0 if no gpu is specified by the -g parameter. This leads to the behavior that crYOLO ignores previous settings in CUDA_VISIBLE_DEVICES. Will be fixed in 1.2.2</​del>​ 
 +  * <​del>​Issue 7: On K3 images crYOLO seems to add a offset toward the longer axis of the input image.</​del>​ 
 +  * <​del>​Issue 8: There is a logical error in filament tracing, which sometimes connects two parallel filaments.</​del>​ 
 +  * <​del>​Issue 9: Some people report an error when running cryolo prediction/​training:​ "​ImportError:​ numpy.core.multiarray failed to import"​. It will be fixed in 1.2.3.</​del>​ 
 +  * <​del>​Issue 10: On machines with many cores (e.g 64) an error during filtering might pop up: "​[ERROR:​0] 53: Can't spawn new thread"</​del>​ 
 +  * <​del>​Issue 11: If the -g parameter is not provided, crYOLO will use the memory of all GPUs. Will be fixed in 1.2.3.</​del>​ 
 +  * <​del>​Issue 12: The LineEnhancer depdenceny of crYOLO is still dependent from opencv. Workaround: In the crYOLO environment:​ conda install ​opencv</​del>​ 
 +  * <​del>​Issue 13: After picking it can happen that some of the boxes are not fully immersed in the image. Will be fixed in 1.2.4.</​del>​ 
 +  * <​del>​Issue 14: Parallelization in filament mode is broken. Will be fixed in 1.2.4.</​del>​ 
 +  * <​del>​Issue 15: If the %%--%%gpu_fraction is used, crYOLO always uses GPU 0Will be fixed in 1.3.1.</​del>​  
 +  * <​del>​Issue 16: %%--%%gpu_fraction only works for prediction, not for training. Will be fixed in 1.3.2.</​del>​ 
 +  * <​del>​Issue 18: Prediction is broken in 1.3.2. It removes all particles as it claim they are not fully immersed in the image.</​del>​ 
 +  * <​del>​Issue 19: Filtering does not work if target image directory is absolute path.</​del>​ 
 +  * <​del>​Issue 20: crYOLO 1.3.4 has a normalization bug. During training the images are normalized seperately, but during prediction is done batch wise. Workaround: Use -pbs 1 during prediction. It will be fixed in 1.3.5.</​del>​ 
 +  * <​del>​Issue 21: The search range for filament tracing is too low for many datasets. To check if you are affected: Use your trained model and pick without the filament options. Check if your filaments a nicely picked (many consecutive boxes on a filament). In the next version, the search range will be increased and added as an optional parameter.</​del>​ 
 +  * <​del>​Issue 22: If absolute paths are used in the field "​train_image"​ in your configuration file, filtering is skipped.</​del>​ 
 +  * <​del>​Issue 23: Since crYOLO 1.4.0 it sometimes take long until it starts picking. The reason seems to be the tensorflow update.<​del>​ 
 +  * <​del>​Issue 24: Fine-tune mode does not start (cannot find layer model_3). Will be fixed in 1.4.1.<​del>​ 
 +  * <​del>​Issue 25: When using GUI, prediction behaves differently than using command line. The reason is, that it uses a different multiprocessing start method. Will be fixed with 1.5.1</​del>​ 
 +  * <​del>​Issue 26: If you select filtering "​None"​ crYOLO does not train properly.</​del>​ 
 +</​hidden>​ 
 +</note>
  
 **That'​s it!** **That'​s it!**
Line 187: Line 191:
 Training crYOLO is much more computational expensive. Training a model with 14 micrographs from scratch on my local machine take 34 minutes per epoch on the CPU. Given that you often need 25 epochs until convergence it is a task to do overnight (~ 12 hours). However, you might want to try [[pipeline:​window:​cryolo##​picking_particles_-_using_the_general_model_refined_for_your_data|refining the general model]], which takes 12 minutes per epoch (~ 5 hours). Training crYOLO is much more computational expensive. Training a model with 14 micrographs from scratch on my local machine take 34 minutes per epoch on the CPU. Given that you often need 25 epochs until convergence it is a task to do overnight (~ 12 hours). However, you might want to try [[pipeline:​window:​cryolo##​picking_particles_-_using_the_general_model_refined_for_your_data|refining the general model]], which takes 12 minutes per epoch (~ 5 hours).
  
-====== ​Start picking! ​======+====== ​Tutorials ​======
  
-Use the **__''​[[pipeline:​window:​cryolo|step-by-step ​tutorial]]''​__** to get started!+Use our **__''​[[pipeline:​window:​cryolo|step-by-step ​tutorials]]''​__** to get started!
  
 ====== Change log ====== ====== Change log ======
  
 ====crYOLO==== ====crYOLO====
 +{{page>​pipeline:​window:​cryolo:​changelog}}
 +====crYOLO Boxmanager====
 +**crYOLO Boxmanager Version 1.2.6:**
 +  * Make it compatible with current new environment
  
-**crYOLO ​1.4.0:** +<​hidden ​**Old crYOLO ​Boxmanager change logs**>
-  * Support Just Another Noise 2 Noise Implemnentation ([[:​janni|JANNI]]) +
-  * Add --mask_width as optional parameter for the filament mode +
-  * Update tensorflow from 1.10.1 to 1.12.3 to make crYOLO compatible with JANNI +
-  * Update numpy from 1.14.5 to 1.15.4 to make crYOLO compatible with JANNI+
  
-<hidden **Old crYOLO change logs**> 
-**crYOLO 1.3.6:** 
-  * Changed filament search radius factor from 0.8 to 1.41 (this fixed issue 21) 
-  * Add search radius factor as [[pipeline:​window:​cryolo#​advanced_parameters|advanced parameter]] (-sr) during prediction in filament mode 
-  * Improved error message in case of corrupted config file 
-  * Fixed issue 22: If absolute paths are used in the field “train_image” in your configuration file, filtering is skipped. 
- 
-**crYOLO 1.3.5:** 
-  * Fixed issue 20: During training the images are normalized separately, but during prediction is done batch wise. The lead to confusing results: some micrographs were perfectly picked, some totally unreasonable,​ even with the same defocus. This bug only affects the picking, already trained models can still be used. 
-  * Remove unnecessary dependencies 
-  * Add %%__%%version%%__%% to %%__%%init%%__%%.py for easy access to package version. 
- 
-**crYOLO 1.3.4:** 
-  * Support for SPHIRE 1.2 
-  * Changed the minimum threshold for cbox files from 0.01 to 0.1. Much faster in many cases but still low enough. If -t is lower than 0.1, the new threshold is used as minimum. 
-  * Installation now checks if python 3 is used. 
-  * Fix issue 19: Filtering does not work if target image directory is absolute path. 
-  * Fix crash when %%--%%otf was specified but filtering was not specified in the config file 
- 
-**crYOLO 1.3.3:** 
-  * Fix issue 18: Prediction is broken in 1.3.2. It removes all particles as it claim they are not fully immersed in the image. 
- 
-**crYOLO 1.3.2:** 
-  * Speedup prediction: Vectorized some parts of the code and optimized the creation of the cbox files. 30% speed up picking / 15% faster training compared to 1.3.1/​1.3.0. ​ 
-  * Bug fix in merging of filaments that sometimes throw "​IndexError:​ list index out of range"​. (Thanks to Alexander Belyy) 
-  * Fix in cryolo_evaluation:​ If the validation data is specified with -b instead of runfiles, all datasets with only one box file were ignored. 
-  * Change library requirement to PILLOW version 6.0.0 
-  * Fix issue 16:  %%--%%gpu_fraction only works for prediction, not for training. 
- 
-**crYOLO 1.3.1:** 
-  * Fix Issue 15: -g was ignored when --gpu_fraction was used. 
- 
-**crYOLO 1.3.0:** 
-  * Fine tune the general network to your data using the new fine tune option with %%--%%fine_tune (https://​1n.pm/​x8rUH) 
-  * One-the-fly micrograph filtering during particle picking with %%--%%otf (don't double your dataset during picking)(https://​1n.pm/​goXAa) 
-  * Interactive threshold adjustment after prediction using the new cbox-files and the crYOLO boxmanager 1.2 (https://​1n.pm/​k7HoI) 
-  * Pick only fully immersed particles (Issue 13) 
-  * Improved filament mode 
-    * Rewrote tracing 
-    * Rewrote and speed up merging of filaments 
-    * Fixed parallelisation of the filament mode (Issue 14) 
-  * Add tifffile as dependency, as imageio throws a lot of warning for some tif files. 
-  * Add conversion for uint16 images, as pillow cannot work with them. 
-  * Add option %%--%%skip_augmentation to deactivate augmentation during training (Thanks to Tijmen de Wolf). (https://​1n.pm/​goXAa) 
-  * Add option %%--%%num_cpu to specify the number of CPUs used during training and during prediction. (Thanks to Nikolaus Dietz) (https://​1n.pm/​goXAa) 
-  * Add option to limit the amount of GPU memory reserved by crYOLO with %%--%%gpu_fraction (Thanks to Nikolaus Dietz) (https://​1n.pm/​goXAa) 
-  * Save anchor size in model every time you write a new model during training (not only at the end) 
-  * In case of using %%--%%min_distance,​ only the particle with lower confidence is removed (Thanks to Yilai Li) 
- 
-**crYOLO Version 1.2.3:** 
- 
-  * crYOLO now saves the anchors which were used during training inside the .h5 file and takes care that the correct anchors are used during prediction. 
-  * [[https://​pypi.org/​project/​lineenhancer|LineEnhancer]] dependency is now installed via PyPi, as --follow-dependency-links is removed in pip 19. 
-  * Fix Issue 9: Removed zignor dependency as it leads to problems for some users (Thanks to Jason Kaelber) 
-  * Attempt to fix Issue 10: Removed opencv dependency which was connected to this problem (Thanks to Shaun Rawson) 
-  * Fix issue 11: crYOLO uses now GPU 0 by default if not specified otherwise (e.g. by CUDA_VISIBLE_DEVICES) 
- 
-**crYOLO Version 1.2.2:** 
-  * Added the PhosaurusNet to the crYOLO backend, which makes the patch mode needless for picking single particles. 
-  * crYOLO now outputs separate folders for EMAN box files and STAR files. ​ 
-  * When picking filaments it will now additionally output EMAN Start-End and STAR Start-End coordinates (Thanks to Jesse M. Hansen). 
-  * Fix Issue 4: The filament mode will crash if crYOLO cannot identify a single particle in the image. 
-  * Fix Issue 5: If movies were aligned with cisTEM and picked with crYOLO, the box positions were vertically flipped. (Thanks to Wei-Chun Kao) 
-  * Fix Issue 6: crYOLO overwrote the CUDA_VISIBLE_DEVICES variable if the -g parameter is not passed. (Thanks to Shaun Rawson) 
-  * Fix Issue 7: crYOLO introduces a shift for non square images proportional to the aspect ratio. (Thanks to Shaun Rawson) 
-  * Fix Issue 8: crYOLO sometimes connects two parallel filaments. The filament tracing was optimized and seems now  working properly. ​ 
-  * Fix a severe bug in filament tracing. Curved filaments splitted by crYOLO in more straight sub pieces. However, during the division, one half of the splitted filament was lost. (Thanks to Sabrina Pospich) 
-  * Added a wiki entry about the [[:​cryolo_nets|networks which are supported by crYOLO]] 
-  * Added a wiki entry about the [[:​cryolo_config|crYOLO configuration file]] 
-  * Added a wiki entry [[:​cryolo_filament_import_relion|how to import crYOLO filament coordinates (from the ''​EMAN_HELIX_SEGMENTED''​ folder) into Relion]]. 
- 
-**crYOLO Version 1.2.1:** 
-  * Fix Issue 2: Tiff files are now written as 32 bit when internal filtering is used. 
-  * cryolo_evaluation now additionally estimates the optimal threshold based on the F2 score, which puts more weight on recall than on precision 
-  * File ending of filament box files is now .box instead of .txt (Thanks to Jesse M. Hansen) 
- 
-**crYOLO Version 1.2.0:** 
-  * Switch to Python3 (**Please use a fresh environment!**) 
-  * (Hopefully) fixed that crYOLO sometimes freezes during/​after training (hard to reproduce, so I'm not 100% sure if it is fixed.) 
-  * Fix that training with multiple GPUs did not speed up small datasets 
-  * Low-pass filtering is now [[http://​sphire.mpg.de/​wiki/​doku.php?​id=pipeline:​window:​cryolo#​picking_-_using_a_model_trained_for_your_data|integrated]] into crYOLO 
-  * Fix two bugs in cryolo_evaluation that lead to an underestimation the performance parameters 
-  * cryolo_evaluation is now multithreaded if your training data is organised in subfolders 
-  * cryolo_evaluation now contains a better method for optimal picking threshold estimation 
-  * Refactoring 
-  * Minor bug fixes 
- 
-**crYOLO Version 1.1.4:** 
-  * Hot fix for filament mode when applied to non square images. ​ 
- 
-**crYOLO Version 1.1.3:** 
-  * Improved non-maximum-suppression brings 60% speedup during picking! 
-  * Multi GPU support for training and prediction (e.g by adding -g 0 1 for GPU 0 and GPU 1 to the training/​prediction command) 
-  * Bug fixed which leads to a crash if no particles are picked on the first micrograph (Thanks to Björn Klink). 
- 
-**crYOLO Version 1.1.2:** 
-  * STAR files could now used for training. However, as they don't contain size information the size specified in the anchors in the config.json is used. 
-  * Slightly improved speed of the filament-mode 
-  * Fixed another bug running filament mode on non-square images (Thanks to Gregory Alushin) 
- 
-**crYOLO Version 1.1.1:** 
-  * More efficient MRC reading and batch prediction leads to ~50% faster training and ~70% faster picking when crYOLO is used in patch-mode (compared with the patch-mode in 1.1.0). 
-  * 6x faster filament picking 
-  * Reading of annotation data is now super-fast :-) (Box filename has to be contained into image filename) 
-  * Optimized filament picking parameters 
-  * Fixed bug which made training fail for some 16 bit images 
-  * Fixed bug which could lead to double picked filaments 
-  * Fixed bug running filament mode on non-square images (Thanks to Gregory Alushin) 
-  * Supports EMAN1 helix coordinates 
-  * Support for star file format. During prediction, both box and star files are written. 
- 
-**crYOLO Version 1.1.0:** 
-  * crYOLO now supports filaments 
-  * New evaluation tool 
-  * Supports empty box files for training on particle-free images 
-  * Extended data augmentation:​ Horizontal flip and flip along both axes 
-  * Experimental support of periodic restarts during training (with --warm_restarts) 
- 
-**crYOLO Version 1.0.4:** 
-  * Fix a problem reading backend weights from read-only filesystem (Thanks to Michael Cianfrocco and Jason Key)  
-  * Make sure that tensorflow version is >= 1.5.0 and < 1.9.0 
-  * Add support for subfolders in training and validation directories 
-  * More clear error message when the trained model does not fit to the architecture specified in the config file. 
- 
-**crYOLO Version 1.0.3:** 
-  * Ignore non-image files during training and predction (Thanks to Kellie Woll)  
-  * Fixed misleading error when non existing folder is used as input for prediction (Thanks to Kellie Woll) 
-  * Add distance threshold during prediction by adding -d distanceInPixel parameter to prediction command (Thanks to Lifei Fu) 
-  * Add "​--write_empty"​ parameter to prediction command if an empty box file should be written if no particle is picked. 
- 
-**crYOLO Version 1.0.2:** 
-  * Fix problem when mrc image has dimensions (1,​width,​height) (Thanks to Reza Behrouzi) ​ 
- 
-**crYOLO Version 1.0.1:** 
-  * Normalization technique is now the same for 8-bit and 32 bit images. 
-  * Unify image augmentation 
- 
-</​hidden>​ 
-====crYOLO Boxmanager==== 
 **crYOLO Boxmanager Version 1.2.3:** **crYOLO Boxmanager Version 1.2.3:**
   * Make it compatible with current new environment   * Make it compatible with current new environment
  
-<hidden **Old crYOLO Boxmanager change logs**> 
 **crYOLO Boxmanager Version 1.2.2:** **crYOLO Boxmanager Version 1.2.2:**
   * Makes sure that the correct version of MatplotLib is used.   * Makes sure that the correct version of MatplotLib is used.
  • downloads/cryolo_1.1562837944.txt.gz
  • Last modified: 2019/07/11 11:39
  • by twagner