Actions
Feature #14113
openptolemy active learning implementation
Start date:
01/11/2023
Due date:
% Done:
0%
Estimated time:
Deliverable:
Description
Using ptolemy_server to decide next square to visit in data collection.
Updated by Anchi Cheng almost 2 years ago
- Assignee changed from @Maria Janssen to Anchi Cheng
Updated by Anchi Cheng over 1 year ago
- Related to Bug #14340: targets not processed in active learning square targeting when mosaic scale is changed added
Updated by Anchi Cheng about 1 year ago
Setting up:
1. Set up ptolemy.cfg based on leginon/ptolemy.cfg.template. The configuration file search pattern follows the same logic as in Configure_leginoncfg.2. Start ptolemy service
3. In leginon environment, test server operation with leginon/ptolemyhandler.py by
python ptolemyhandler.py
- If it can not find the hardcoded MosaicTileData dbid=1945 in your leginondb database, please change it at the end of the script to one that is valid for the test.
4. Replace the two nodes for "Square Targeting" and "Exposure Targeting" with node classes that uses active learning ptolemy in your MSI application. See Use_the_Application_Editor_to_create_Leginon_applications on how to edit using gui
"Square Targeting": MosaicLearnTargetFinder
"Exposure Targeting": PtolemyMmTargetFinder
5. Launch leginon and your modified MSI application, set settings of these new nodes.
- The settings for MosaicLearnTargetFinder is the same as MosaicScoreTargetFinder class
- The settings for PtolemyMmTargetFinder is similar to ScoreTargetFinder but without the need to specify the script path
Actions