6th week (July 1 - July 7)
1. Implemented LazySnapping without watershed filtering. The implementation doesn't quite work yet.
At the moment I'm using a naive implementation of K-means. But I'm thinking of using the implementation provided at http://www.cs.umd.edu/~mount/Projects/KMeans/ (release under GPL)
2. add option for choosing segmentation technique
Initially basic polygon editing is chosen. This can be changed from Edit->Preferences. The segmentation technique can be changed anytime. However the previous masks (if any) are discarded. Later on I'll provide option for reusing previous mask.
Not done yet
1. Working implementation of LazySnapping.
2. Did not make use of watershed filtering. I'm thinking of just using watershed.cxx code which is part of VIGRA examples. The only thing I'll need to do is build a graph from the output of the watershed algorithm.
Plan for second phase:
For the second phase I'm planning to give high priority to integration and usability (eg. fast response, easily going back and forth between different images that are edited) and slightly lower priority to fine-tuning, and specialized mask creation. Plan for the next three weeks is as follows -
7th week (July 8 - July 14) Midterm evaluation week
1. Apply segmentation on the overlapped region
2. work on preview window (eg. show images that are selected and not selected)
This is basically giving a preview of loaded images. I'll keep it very simple for now.
3. Select multiple images for display
4. Work on LazySnapping. I'll mainly be focusing on the part without watershed filtering.
5. Determine which image is being edited
8th Week (July 15 - July 21)
2. Start working on integration with Hugin
3. Start working on fine-tuning mask in polygon mode
4. Do further improvement of lazysnapping.
9th Week (July 22 - July 28)
1. Undo/Redo feature
2. Load/Store mask
3. Further work on integration (I'll determine the deliverables on 8th week)