This code implements the algorithms described in "Image Provenance at Scale", with respect to the step of "graph building" (right side of the illustration above).
- Python 2.7
- numpy
- opencv 2.4
- src Source content. Auxiliary libs are stored in "lib". Main programs are numbered in the way they might be executed for a complete provenance graph building:
- 01_build_adj_mat.py: Builds the adjacency (dissimilarity) matrices and saves them in a proper ".npz" file.
- 02_join_adj_mat.py: Joins/combines the adjacency matrices that are saved in a given ".npz" file.
- 03_build_graph.py: Builds the provenance DAG from the given adjacency matrix.
- script
NIST-Nimble-based execution shell scripts. The scripts are numbered in the way they might be executed for a complete provenance graph building.
Please contact Aparna Bharati ([email protected]) or Daniel Moreira ([email protected]).