User Tools

Site Tools


cryolo_train_general_model

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
Last revision Both sides next revision
cryolo_train_general_model [2019/04/19 16:49]
twagner
cryolo_train_general_model [2019/04/19 16:54]
twagner
Line 3: Line 3:
 Training a model for a specific dataset is very easy with crYOLO. However, you might have multiple data collections of the same particle with different settings, a different camera or another microscope. A model trained on the data of one data collection, might not perform very good on a dataset from another data collection. Training a model for a specific dataset is very easy with crYOLO. However, you might have multiple data collections of the same particle with different settings, a different camera or another microscope. A model trained on the data of one data collection, might not perform very good on a dataset from another data collection.
  
-However, you can easily train a crYOLO model that **generalize** well on data recorded under multiple conditions. To achieve this, all you have to do is to **merge training data** of multiple datasets. The result will be a model that be applied to new dataset from a new data collection without additional training. +However, you can easily train a crYOLO model that **generalize** well on data recorded under multiple conditions. To achieve this, all you have to do is to **merge training data** of multiple datasets. The result will be a model that can be applied to new dataset from a new data collection without additional training. 
  
 Here is our recommendation how to organize the training data. Instead of copying your images and box files directly into ''train_images'' / ''train_annot'', you can copy them into subfolders. One for each data collection: Here is our recommendation how to organize the training data. Instead of copying your images and box files directly into ''train_images'' / ''train_annot'', you can copy them into subfolders. One for each data collection:
Line 10: Line 10:
  
 The //train_image_folder// and //train_annot_folder// parameters in the [[cryolo_config|crYOLO configuration file (e.g. config.json)]] still point to the root directories ''train_images'' and ''train_annot'' respectively. The parameter //anchors// should be set roughly to mean of all particle box sizes. Other than that, the training of a general model does not differ from [[http://sphire.mpg.de/wiki/doku.php?id=pipeline:window:cryolo#picking_particles_-_using_a_model_trained_for_your_data|training a model from scratch]]. The //train_image_folder// and //train_annot_folder// parameters in the [[cryolo_config|crYOLO configuration file (e.g. config.json)]] still point to the root directories ''train_images'' and ''train_annot'' respectively. The parameter //anchors// should be set roughly to mean of all particle box sizes. Other than that, the training of a general model does not differ from [[http://sphire.mpg.de/wiki/doku.php?id=pipeline:window:cryolo#picking_particles_-_using_a_model_trained_for_your_data|training a model from scratch]].
 +
 +One minor hint: When running the training of a general model, we always use [[http://sphire.mpg.de/wiki/doku.php?id=pipeline:window:cryolo#advanced_parameters|--warm_restarts]].
cryolo_train_general_model.txt ยท Last modified: 2019/04/19 17:53 by twagner