User Tools

Site Tools

cryolo_reference_example

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
Next revision Both sides next revision
cryolo_reference_example [2019/04/03 15:53]
twagner [crYOLO reference example] 		cryolo_reference_example [2019/09/23 09:27]
twagner [Prediction]
Line 4: Line 4:
  
 ===== Reference setup ===== ===== Reference setup =====
-We run this example on machine with the following specification:+We run this example on a machine with the following specification:
   * Titan V   * Titan V
   * Intel Core i9 7920X @ 2.90 Ghz   * Intel Core i9 7920X @ 2.90 Ghz
 +  * SSD Harddrive
  
  
Line 22: Line 23:
  
 The ''toxin_reference'' directory contains multiple folders / files: The ''toxin_reference'' directory contains multiple folders / files:
-  * train_image: Folder with the training images +  * train_image: Folder with 12 training images 
-  * train_annot: Folder with box files for the training images+  * train_annot: Folder with 12 box files for the training images
   * config_phosnet.json: Configuration file for crYOLO   * config_phosnet.json: Configuration file for crYOLO
   * reference_model.h5: Model that I've trained on my machine using the commands below.   * reference_model.h5: Model that I've trained on my machine using the commands below.
Line 35: Line 36:
 ===== Training ===== ===== Training =====
  
-First, you have to warm up the network with "-g 0" I'm selection GPU 0. It takes 2 minutes 30 seconds on our reference setup: + 
-<code> +The training is done with this command:
-cryolo_train.py -c config_phosnet.json -w 5 -g 0 +
-</code> +
-The actual training is done with this command:+
 <code> <code>
-cryolo_train.py -c config_phosnet.json -w 0 -e 10 -g 0+cryolo_train.py -c config_phosnet.json -w 5 -e 5 -g 0
 </code> </code>
-It needed 6 minutes 45 seconds to converge (18 epochs). The best validation loss was 0.03316. These numbers might be a little bit different on your case.+It needes 5 minutes 50 seconds to converge (5 warmup + 10 "normal" epochs). The best validation loss was 0.03042. These numbers might be a little bit different on your case.
  
 ===== Prediction ===== ===== Prediction =====
Line 50: Line 48:
 </code> </code>
  
-It picked 1617 particles on 12 micrographs in 4 seconds. Including filtering the image and loading the model the command needed 46 seconds.+It picked 1617 particles on 12 micrographs in 3 seconds. Including filtering the image and loading the model the command needed 38 seconds.
  
 ===== Visualize results ===== ===== Visualize results =====
cryolo_reference_example.txt · Last modified: 2020/05/08 11:43 by twagner

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: GNU Free Documentation License 1.3
GNU Free Documentation License 1.3 Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki