1.8.3

What's Changed

• Fix: Update PORT_CONNECT variable to 5555 for consistent server confi… by @Moobbot in #16

Full Changelog: v1.8.0...1.8.3

v1.8.0

What's Changed

Major Improvements

1. Configuration Management

• Added new configuration variables HOST_CONNECT and PORT_CONNECT for improved server settings flexibility
• Enhanced file naming logic in call_model.py to support dynamic file types based on configuration

2. Performance & Processing

• Improved attention score processing with reversed indexing for predicted filenames
• Enhanced image ranking logic with better intensity metrics
• Streamlined attention score calculations in rank_images_by_attention function

3. Code Organization & Architecture

• Restructured import statements and enhanced function documentation
• Introduced new visualization utility module for attention overlay and image saving
• Improved code organization and readability in call_model.py

4. Error Handling & Logging

• Enhanced error handling in the prediction workflow
• Improved logging for DICOM file-saving operations
• Better error reporting in overlay image handling

5. Visualization & Output

• Refined overlay image handling in routes.py
• Updated visualization configuration usage for consistency
• Enhanced attention overlay visualization

6. API & Integration

• Added API endpoints for result upload
• Improved server connectivity and configuration
• Enhanced DICOM metadata handling

Technical Details

• Updated default attention threshold for improved numerical stability
• Streamlined visualization process
• Enhanced documentation and code clarity
• Improved error handling and logging throughout the application
• Better organization of code structure and modularity

These changes represent significant code quality, performance, and maintainability improvements while adding new features and enhancing existing functionality.

Full Changelog: v1.7.1-alpha...v1.8.0

v1.7.1-alpha

What's Changed

• Api custom by @Moobbot in #1
• Set up virtual environment in Dockerfile and simplify PyTorch installation in setup.py by @Moobbot in #2
• Update Dockerfile to create a virtual environment, upgrade pip, and i… by @Moobbot in #3

API - Docker custom

• Configured Flask application with upload and results directories.
• Implemented functions to handle file uploads, download checkpoints, clean up old results, save uploaded files, load models, and predict using the Sybil model.
• Added API endpoints for prediction, file download, file preview, GIF download, and DICOM to PNG conversion.
• Improved logging and error handling throughout the application.
• Enhanced prediction function to support DICOM and PNG file types, and added options for saving predictions as DICOM images.
• Introduced session management using UUIDs for each prediction request.
• Added functionality to visualize attention maps and save them as overlay images.
• Implemented local IP address retrieval for better network configuration.
• Updated application to run on all IP addresses, including localhost and local network.

Full Changelog: v1.6.0...v1.7.1-alpha