Salps

Repository for scripts to run salp poop segmenting and analyze average color from masked regions.

Overview

This project contains scripts for analyzing salp poop color characteristics from segmented images. The main script mask_color.py processes binary mask files and calculates color statistics for the masked regions.

Prerequisites

• Python 3.6+
• Required packages (install via pip):

  pip install opencv-python numpy pandas tqdm

Configuration

Important: Update Base Directory

Before running the script, you must update the base_dir variable (I hard-coded it..) in scripts/mask_color.py:

# Path to the project directory
base_dir = "/path/to/your/salp/project"  # CHANGE THIS TO YOUR PATH

The current path is hardcoded and will not work on other systems.

Running the Script

Input Requirements

The script expects the following directory structure:

your_project_directory/
├── session_*/                    # Session directories
│   └── filled_masks/            # Contains binary mask files
│       └── *_mask.png           # Mask files (white = poop, black = background)
├── OFP*/                        # Original image directories
│   └── *.JPG                    # Original images (must match mask names)

Running the script

Just run the script with this command (make sure you're at the right level of the directory structure):

python3 mask_color.py

Outputs

The script generates several types of output:

1. Output masks as a sanity check (On the first 3 processed images)

• Location: debug_masked_areas/ directory
• Files:
  • debug_*_overlay.png - Original image with red overlay on masked regions
  • debug_*_diff.png - Difference images (if comparing masks)

2. Color Analysis CSV Files

• Location: session_*/salp_poop_color_summary.csv
• Contents:
  • Session name
  • Mask and image filenames
  • Average RGB color values
  • Calculated brightness
  • Color category (Dark, Light, Reddish, Greenish, Bluish, Mixed)

3. Output on the command line

• Processing progress with tqdm
• Debug information for first 3 images
• Summary statistics including:
  • Brightness range and averages
  • Color category distribution
  • Most similar and most different color pairs
  • Session-wise analysis

Color Analysis Features

Brightness Calculation

Uses ITU-R Recommendation BT.601 standard:

Brightness = 0.299 * R + 0.587 * G + 0.114 * B

Color Categorization

Simple RGB dominance-based categories:

• Dark: All RGB values < 50
• Light: All RGB values > 200
• Reddish: R > G and R > B
• Greenish: G > R and G > B
• Bluish: B > R and B > G
• Mixed: No clear dominance

Color Similarity Analysis

• Calculates Euclidean distance between all color pairs
• Identifies most similar and most different colors
• Provides distance metrics for comparison

Troubleshooting

Common Issues

1. "Missing image for [filename]"

   • Ensure original images exist in OFP* directories
   • Check that image names match mask names (replace _mask.png with .JPG)

2. "Error reading [file]"

   • Verify file paths are correct
   • Check file permissions
   • Ensure images are in supported formats (JPG, PNG)

3. "Mask is not binary"

   • Masks should contain only 2 unique values (typically 0 and 255)
   • Check mask generation process

Debug Mode

The script automatically generates debug information for the first 3 processed images, including:

• Mask analysis (shape, data type, value range)
• Pixel count statistics
• Overlay images for visual verification

File Structure Example

your_project_directory/
├── session_001/
│   ├── filled_masks/
│   │   ├── image001_mask.png
│   │   └── image002_mask.png
│   └── salp_poop_color_summary.csv
├── session_002/
│   ├── filled_masks/
│   │   └── image003_mask.png
│   └── salp_poop_color_summary.csv
├── OFP_2023_01/
│   ├── image001.JPG
│   ├── image002.JPG
│   └── image003.JPG
└── debug_masked_areas/
    ├── debug_image001_overlay.png
    └── debug_image001_diff.png