Differences

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

--- auto_2d_class_selection [2019/07/11 14:03]
twagner
+++ auto_2d_class_selection [2020/08/27 15:11] (current)
twagner [Changelog]
@@ Line 1: / Line 1: @@
-====== Cinderella: Automatic 2D class selection ======
+====== Cinderella: Deep learning based binary classification tool ======
 ----
@@ Line 8: / Line 8: @@
 ----
+Our binary classification  tool (//Cinderella//) is based on a deep learning network to classify class averages, micrographs or subtomograms into good and bad categories.
+Cinderella supports ''.hdf/.mrcs'' ** files for class averages**, ''.mrc'' **files for micrographs**, and ''.hdf'' **files for subtomograms**.
+//Cinderella// was written to automate cryo-em data processing.  It's open source and easy to use.
+We provide a pretrained general model for classifying class averages ([[auto2d_tutorial|see tutorial]]). But you can easily train it with your own set of classes, micrographs, and/or subtomograms.
-Our automatic 2d class selection tool (**Cinderella**) is based on a deep learning network to seperate 2D classes from .hdf / .mrcs files into good and bad classes. It uses the same deep neural network as crYOLO and was pretrained on a set good / bad classes. Cinderella was written to automate cryo-em data processing. It's open source and easy to use ([[auto2d_tutorial|see tutorial]]). You can easily train it with your own set of classes.
 <note>
-  * **License**: MIT
+  * **License**: [[https://github.com/MPI-Dortmund/sphire_classes_autoselect/blob/master/LICENSE|MIT]]
-  * **GitHub repository**: https://github.com/MPI-Dortmund/sphire_classes_autoselect
+  * **Repository**: [[https://github.com/MPI-Dortmund/sphire_classes_autoselect|GitHub]]
 </note>
-Here are a couple of examples for good / bad classes in Cinderella:
+====== 2D class selection model ======
+Our model was trained on a set of 2D classes from both [[https://sphire.mpg.de/wiki/doku.php?id=pipeline:isac:sxisac2|ISAC]] and Relion. During the creation of the training data set, we tried to answer the question, "Which class would I select If I would not know the particle?" when deciding what is a "good" class. Here are a couple of examples for good/bad classes in //Cinderella//:
 {{ ::cinderellea.png?450 |}}
-====== The Model ======
-Our model was trained on a set of 2D classes from ISAC. During the creation of the training dataset, I tried to ask myself "Which class would I select If I would not know the particle?" to decide which is a good class.
-<note important>
-The training dataset does **not contain any Relion classes**, so it might be that Cinderella will not work with well them.
-</note>
 You can easily [[auto_2d_class_selection#contribute|contribute]] your own classes!
-Right now our model is trained on **19 datasets**. But we will increase the number often!
+Right now our model is trained on **4773 good classes and 5390 bad classes**.
 ====== Download ======
 ====Cinderella====
-Version: 0.3.0
+Version: 0.7.0
-Uploaded: 11. July 2019
+Uploaded: 27. August 2020
 [[https://pypi.org/project/cinderella/#files|DOWNLOAD]]
-====Pretrained model====
+====Pretrained model (2D classes)====
-Uploaded: 11. July 2019, Datasets: 19
+Uploaded: 27. August 2020, Dataset: 4773 good classes and 5390 bad classes.
-[[ftp://ftp.gwdg.de/pub/misc/sphire/auto2d_models/model_cinderella_20190708.h5|DOWNLOAD]]
-[[auto2d_tutorial#classify|Valid configuration file]]
+[[ftp://ftp.gwdg.de/pub/misc/sphire/auto2d_models/gmodel_cinderella07_202008_N10163.h5|DOWNLOAD]]
 ====Archive====
-Old version of cinderella and the pretrained model can be found in the [[cinderella_archive|archive]]
+Old versions of cinderella and the pretrained model can be found in the [[cinderella_archive|archive]].
 ====Changelog====
+=== Version 0.7 ===
+  * Now uses a **circular masks by default**. This allows to use full rotation during data augmentation. Can be deactivated by setting the field ''mask_radius'' in the configuration file to -1. In case you want to use an model trained with Cinderella < 0.7 please set the radius to -1. Otherwise you case specify any radius you want. **By default** (no ''mask_radius'' provided) it will use 0.4*''input_size''.
+  * The general models now includes **300 new good Relion classes and 2000 new bad Relion classes** (//Thanks to Takanori Nakane and Grigory Sharov//).
+  * Fixed numerical instability that occurs when you have classes filled with a constant value (//Thanks to Grigory Sharov//).
+  * Fixed a problem with classes that contain NaN values. NaN values are now replaced with 0. (//Thanks to Grigory Sharov//).
+  * Fixed an issue when filenames contain more than one point.
+=== Version 0.6 ===
+  * Fix an issue for classes in mrcs format
+  * Minor changes
+=== Version 0.5 ===
+  * Add support for subtomograms
+  * Faster file reading
+=== Version 0.4 ===
+  * Balances unbalanced training datasets.
+  * It is now possible to train Cinderella to select micrographs
+  * Updated the general model for 2D class selection.
+=== Version 0.3.1 ===
+  * Downgrade to tensorflow 1.10.1 again, as user report long initialization times
+  * Only report the number of good / bad classes + their fraction.
 === Version 0.3.0 ===
   * More data augmentation (add rotation)
@@ Line 69: / Line 93: @@
 After that, create a new virtual environment:
 <code>
-conda create -n cinderella -c anaconda python=3.6 pyqt=5 cudnn=7.1.2
+conda create -n cinderella -c anaconda python=3.6 pyqt=5 cudnn=7.1.2 numpy==1.14.5
 </code>
@@ Line 75: / Line 99: @@
 <code>
 source activate cinderella
-</code>
-Install fast numpy:
-<code>
-conda install numpy==1.15.4
 </code>
@@ Line 93: / Line 112: @@
 ====== Tutorial ======
-[[auto2d_tutorial|We created a tutorial how to use Cinderella!]]
+We created three tutorials:
+  * [[auto2d_tutorial|How to use Cinderella for 2D class selection]]
+  * [[cinderella_micrographs|How to use Cinderella for micrograph selection]]
+  * [[cinderella_tomograms|How to use Cinderella for subtomogram selection]]

User Tools

Site Tools

Differences

Page Tools