Interactive GUI application for extracting individual photos from scanned images.

If you've scanned multiple physical photos on a flatbed scanner in one pass, scan-crop will automatically detect each photo and let you adjust the crop regions, apply rotations, and export them as separate high-quality JPEG files.

• Automatic photo detection using computer vision
• Interactive crop adjustment with mouse drag and resize
• Manual region creation by dragging on the canvas
• Rotation controls (90° increments) for each photo
• Live preview of selected photo with transformations
• Keyboard shortcuts for power users
• Batch export of all photos at once
• Command-line support for drag-and-drop workflow

No Python installation required!

1. Download the latest scan-crop.exe from Releases
2. Run it directly - no installation needed

Requirements: Python 3.11 or higher

1. Clone this repository:

git clone https://github.com/rusanu/scan-crop.git
cd scan-crop

2. Create a virtual environment and install dependencies:

python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt

3. Run the application:

python scan_crop_gui.py

Using the standalone executable:

• Double-click scan-crop.exe to launch
• Or drag a JPEG image file onto scan-crop.exe
• Or run from command line: scan-crop.exe path\to\scan.jpg

Running from source:

python scan_crop_gui.py              # Launch with file browser
python scan_crop_gui.py scan.jpg     # Open specific image

1. Load an image - Click "Open Image..." or launch with a file path
2. Review detected regions - Green boxes show automatically detected photos
3. Adjust crop regions - Click and drag regions to move, drag corners/edges to resize
4. Create manual regions - Drag on empty areas to create new crop regions
5. Select a region - Click any region to select it (shows in right preview panel)
6. Rotate photos - Use toolbar buttons or click region and use rotation controls
7. Delete regions - Select unwanted regions and click "Delete Region" or press Delete
8. Export all - Click "Export All Photos..." to save all regions to a folder

• Tab / Shift+Tab - Cycle through detected regions
• Arrow keys - Move selected region (1 pixel)
• Shift+Arrow keys - Resize selected region (1 pixel)
• Delete - Delete selected region

For the complete shortcuts list, see Help → Keyboard Shortcuts in the application.

1. Scan Quality: Use 300 DPI or higher for best detection
2. Spacing: Leave space between photos on the scanner bed
3. Contrast: Ensure good contrast between photos and scanner background
4. Lighting: Avoid shadows and reflections on the scanner glass
5. Orientation: Photos can be in any orientation - adjust rotation in the app

scan-crop uses OpenCV computer vision to:

1. Convert the scanned image to grayscale
2. Apply binary thresholding using Otsu's method
3. Clean up noise with morphological operations
4. Find contours (outlines) of objects in the image
5. Filter contours by size, aspect ratio, and dimensions
6. Present detected regions in an interactive GUI for review and adjustment
7. Apply rotations and export each photo as a high-quality JPEG (95% quality)

If you encounter errors related to OpenCV DLLs (especially "Invalid number of channels" or similar errors), the application includes diagnostic logging to help identify the issue.

To generate diagnostic logs:

1. Run the application with the --debug flag:

   scan-crop.exe --debug path\to\image.jpg

2. A log file will be created in the same directory as the executable: scan-crop-error-YYYYMMDD.log

3. The log file contains:

   • OpenCV version and installation path
   • DLL files being used
   • Python environment details
   • System PATH information
   • Full error stack traces

4. If the application crashes, diagnostic information is automatically logged

Common causes:

• Conflicting OpenCV installations on your system
• Missing or incompatible DLL files
• System environment variables pointing to wrong OpenCV versions

Solutions:

• Try running from a clean directory without other OpenCV installations in PATH
• Check the log file for DLL paths and ensure they match the packaged version
• Report the log file contents when filing a bug report

This project is free to use and modify.