CV_Project

Shadow Detection & Removal from Images

How to Run?

• Install Python 3.10 (if not 3.10 ensure 3.8 atleast)
• Go to desired ipynb file, for eg. final.ipynb
• Run all cells

How to modify Hyperparameters?

In the Shadow Remover class all the hyperparameters used in the three techniques provided are taken as input while we're initializing the class.

class ShadowRemover:
        itr: # Number of Iterations the algorithm will run
        method: # Technique to be used for mask calculation
        lab_adjustment: # Boolean for using Lab Adjustment for removal
        ab_threshold: # The AB-Threshold for Binarization during mask calculation
        region_adjustment_kernel_size: # Region Adjustment kernel used for Shadow Removal
        shadow_dilation_kernel_size: # Kernel size for highlighting shadows while removal
        shadow_dilation_iteration: # Number of Iterations for Shadow Dilation
        shadow_size_threshold: # The Threshold for detecting shadow sizes
        verbose: # For displaying every region in every iterations
        plot_results: # For displaying the results in plot

Directory Structure

• Experiments - All things we tried
  • colour_adjustment.py - Can modify pixel colours
  • model.ipynb - For running the DL Model
  • scanned_documents.ipynb - Shadow Removal on Scanned Documents
  • trained_model4.pth - Trained model weights
• Images - Folder for images
• Results - Result Images
• final.ipynb - Final ipynb file containing the code of our final implementation

Contributors

Name	Year	Department
Anadi Sharma	B21CS008	Computer Science and Engineering
Drishti Agrawal	B21CS027	Computer Science and Engineering
Shubh Goyal	B21CS073	Computer Science and Engineering
Sukriti Goyal	B21CS075	Computer Science and Engineering