Task board

Task board for Computer Vision project!

Basic SWT Implementation

• Complete code for splitting words

Arbitrary angle detection

• Reading and understanding the paper

Utilities for advanced algorithm

• CAMShift algorithm
• Random tree training
• Ground truth extraction module
• Hierarchical agglomerative clustering
• Color self-similarities
• Contour shape
• Edge shape
• Component level features
• Chain level features
• Final evaluation metrics
• Modify candidate linking
• Modify component analysis
• Training data

To be done

• Improving OCR (vishwa)
• Chain level training (satwik)
• Ground truth size issue (vishwa/satwik)
• Embed random forest classifier into the pipeline (satwik)
• Computing the tightest bounding box for the chain (checking left)
• Quantitative evaluation vanilla (vishwa)
• Quantitative evaluation advanced (vishwa)
• Tuning the parameters (vishwa/satwik)
• Poster (vishwa/satwik)

OCR (optional)

• Look at methods that can be used from 3rd party sources

Final Wrap up

• Clean up all the code files, include inline comments
• Document the project explaining what each function / module does
• Write up the wiki page for the whole project and wrap up everything