Deliverable #4: Improved Paikin & Tal Solver
with Integrated Accuracy Metrics
This deliverable is an improved version of the program that was submitted as deliverable #3. As with the previous deliverable, the set of required Python packages is unchanged. They are:
- OpenCV: An open-source computer vision library
- NumPy: A package that simplifies large matrix computations.
- Pickle: An object serialization library
If you are running Python in Windows (tested in Windows 7 and 10 only), you can use the Miniconda Python Distribution; note, Miniconda is the Python platform I used for my development. Windows users can download a Windows batch file to install all of the needed packages. An archived version of the documentation associated with this deliverable is also available.
New features available in this release are listed below. For definitions and explanation of key terms, please see the first semester final report.
- Enhanced Direct Accuracy Score (EDAS) Calculation
- Shiftable Enhanced Direct Accuracy Score (SEDAS) Calculation
- Enhanced Neighbor Accuracy Score (ENAS) Calculation
- Visualization Tools for EDAS, SEDAS, and ENAS
- Best Buddy Accuracy Calculation
- Standalone Image Best Buddy Analyzer - This is accessed via the module "best_buddy_image_analyzer.py"
- An Initial Best Buddy Placer
|