---
license: apache-2.0
tags:
- skin
- disease
- health
---

### Key Features

+ Images are stored as binary data (more efficient than base64)
+ Standard 224×224 resizing for ML compatibility
+ Automatic RGB conversion for consistency
+ Both original and processed dimensions tracked

### Data Structure:

+ `disease_label`: disease_name,  # String - Disease classification/name
+ `image_filename`: image_file.name,  # String - Original filename
+ `image_path`: str(image_file),  # String - Full path to original image
+ `image_data`: img_bytes,  # Bytes - Raw JPEG image bytes
+ `original_width`: original_width,  # Int - Source image width in pixels
+ `original_height`: original_height,  # Int - Source image height in pixels
+ `processed_width`: img.width,  # Int - Final processed image width
+ `processed_height`: img.height,  # Int - Final processed image height
+ `channels`: len(img.getbands()),  # Int - Number of color channels (3 for RGB)
+ `file_size_bytes`: image_file.stat().st_size,  # Int - Original file size
+ `compressed_size_bytes`: len(img_bytes)  # Int - Compressed JPEG size

### Copyright

(c) Copyright 2025 Finbarrs Oketunji. All Rights Reserved.