https://github.com/tianzonglin/cloud-control-gui
A tool to compute, visualize, analyse and drag points (high-dimensional data)
https://github.com/tianzonglin/cloud-control-gui
cuda interaction-design visualization
Last synced: about 2 months ago
JSON representation
A tool to compute, visualize, analyse and drag points (high-dimensional data)
- Host: GitHub
- URL: https://github.com/tianzonglin/cloud-control-gui
- Owner: TianZonglin
- Created: 2020-07-30T20:06:32.000Z (almost 6 years ago)
- Default Branch: master
- Last Pushed: 2020-10-13T17:59:32.000Z (over 5 years ago)
- Last Synced: 2025-05-22T07:12:50.718Z (about 1 year ago)
- Topics: cuda, interaction-design, visualization
- Language: C++
- Homepage:
- Size: 13.8 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
BUILDING
========
The software was tested to build and run under Linux Ubuntu. As a prerequisite, CUDA 4.0 or higher should be installed.
To build:
1. Go to LIBRARIES/glui-master. Possibly re-run CMake to update to your platform's configuration.
2. Do a "make" from the root directory
To clean everything:
1. Do a "make realclean" from the root directory.
RUNNING
=======
Example:
projwiz -f DATA/segmentation lamp
DONE
====
* Better selection mechanism. We can now select points and groups-of-points. Selection is additive and can be reset. All selections work by clicking in the main window:
-normal click: select closest point to mouse;
-CTRL-click: select entire (label) group under the mouse;
-SHIFT-click: add points to selection rather than overwriting it; works in both normal and CTRL modes;
-click far away from any point: clear selection;
* False-negative bundles: Now they're done for either the entire dataset or the current selection:
-the current selection is void: FN's are shown for the entire dataset;
-the current selection is not void: only FN's of points in the selection are used;
* False-negative map: The map is now computed w.r.t. the current selection. That is:
-if the crt-selection is 1 point: same result as the original idea (show FN-error for all other points w.r.t. selected point)
-if the crt-selection is more points: for all points p not in selection, show _minimal_ FN-error w.r.t. _all_ points in the selection.
TODO
====
* Add possibility to select also groups from the visual clustering; for this, we must detect connected-components in the visual clustering.
* Add generic way to show min/max/avg of the various plotted signals (FNs, FPs, avg-error, etc)
* Show spatial-difference/error-difference between 2 projections