SPHIRE Workflow
Sidebar
pipeline:window:cryolo
This version is outdated by a newer approved version.
This version (2020/05/25 10:12) was approved by twagner.The Previously approved version (2020/02/13 14:02) is available.
This version (2020/05/25 10:12) was approved by twagner.The Previously approved version (2020/02/13 14:02) is available.This is an old revision of the document!
Table of Contents
Overview
NEW DOCUMENTATION
The documentation has moved to https://cryolo.readthedocs.io/en/latest/
CrYOLO is a fast and accurate particle picking procedure. It's based on convolutional neural networks and utilizes the popular You Only Look Once (YOLO) object detection system.
- crYOLO makes picking fast – On a modern GPU it will pick your particles at up to 6 micrographs per second.
- crYOLO makes picking smart – The network learns the context of particles (e.g. not to pick particles on carbon or within ice contamination )
- crYOLO makes training easy – You might use a general network model and skip training completely. However, if the general model doesn't give you satisfactory results or if you would like to improve them, you might want to train a specialized model specific for your data set by selecting particles (no selection of negative examples necessary) on a small number of micrographs.
- crYOLO makes training tolerant – Don't worry if you miss quite a lot particles during creation of your training set. crYOLO will still do the job.
In this tutorial we explain our recommended configurations for single particle and filament projects. You can find more information how to use crYOLO, about supported networks and about the config file in the following articles:
You can find more technical details in our paper:
We are also proud that crYOLO was recommended by F1000:
“CrYOLO works amazingly well in identifying the true particles and distinguishing them from other high-contrast features. Thus, crYOLO provides a fast, automated tool, which gives similar reliable results as careful manual selection and outperforms template based selection procedures.”
Bettina Böttcher, Biochemistry, University WürzburgInstallation
You can find the download and installation instructions here: Download and Installation
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.
Release notes
DOCUMENTATION OUTDATED
Find new versions here: https://cryolo.readthedocs.io
crYOLO 1.6.1:
- Fixed a bug that was introduced with 1.5.5: Scaling of the anchor boxes was wrong. This leads to longer and unstable training and heavily affects the fine-tune mode. (Thanks to Jorge Jimenez de la Morena and Pablo Conesa)
- Fixed a bug that leads to an exception (_tkinter.TclError: couldn't connect to display) at the end of the training on cluster machines. (Thanks to Wolfgang Lugmayr)
crYOLO 1.6.0:
- In case of the general model, you can specify with --minsize MIN --maxsize MAX a minimum and maximum size. This will filter the particles according to their estimated size.
- The estimated size and confidence distribution are now written in a new subfolder DISTR in your output folder. It will also write .csv files with a summary of the distributions.
- In case of the general model, you don't need to specify the anchor size anymore.
- With every run, crYOLO now writes the command used into the central log directory.
- All log files (runfiles, commands, tensorflow) are now saved in the central log directory.
- During training, the intermediate models now get a suffix “_tmp”. After training is finished they are renamed to the specified name in the configuration file (field: “save_weights_name”).
- The boxmanager can now be started through the crYOLO GUI.
- Fixed issue that the filament mode does not work with micrographs that were motion-corrected by unblur.
- Fixed issue that the flaq --write_empty did not work for the filament mode.
- Fixed issue that the minimum distance filter was not applied on particles in .cbox files.
- Fixed issue with the evaluation tool that crashed if no particle can be found for a specific threshold.
(Box filename has to be contained into image filename)Tutorials
Depending what you want to do, you can follow one of these self-contained Tutorials:
The first, second and third tutorial are the most common use cases and well tested. The fourth tutorial is still experimental but might give you better results in less time and less training data.
Help
Any questions? Problems? Suggestions?
Find help at our mailing list!
pipeline/window/cryolo.1590394356.txt.gz · Last modified: 2020/05/25 10:12 by twagner
Page Tools
Except where otherwise noted, content on this wiki is licensed under the following license: GNU Free Documentation License 1.3
