How meyraj/bulk-image-upscaling-project works
The product is an image super-resolution application with batch and watched-folder entry points. Its main workflow discovers and validates images, loads selected model weights, performs single-model or ensemble inference, and saves upscaled results. The repository also includes channel, PSNR, and sharpness diagnostics. Edges reflect the sampled calls and explicit README descriptions; the diagnostics are separate from the production inference path.
Main entry points: inference.py Batch inference · watch.py Folder watcher · model_loader.py Model loading · verify_channels.py Channel verification
Components 18
Processing entry points
- Folder watcher
image_upscaler/watch.py - Batch inference
image_upscaler/inference.py
Image processing
- Image discovery
image_upscaler/src/image_loader.py - Single upscaler
image_upscaler/src/upscaler.py - Ensemble upscaler
image_upscaler/src/upscaler.py - Model loading
image_upscaler/src/model_loader.py
Runtime support
- Model weights
image_upscaler/models - Processing settings
image_upscaler/src/config.py - Application logging
image_upscaler/src/logger.py - Directory and timing helpers
image_upscaler/src/utils.py - Input images
image_upscaler/input - Upscaled images
image_upscaler/output - Processed archive
image_upscaler/input
Quality diagnostics
- Channel verification
verify_channels.py - Comparison images
- PSNR comparison
check_psnr.py - Sharpness analysis
measure_sharpness.py
Other components
- User
Connections 25
- User → Batch inference: runs batch
README.md - User → Folder watcher: starts watcher
README.md - Folder watcher → Batch inference: launches inference
image_upscaler/watch.py - Folder watcher → Input images: scans files
image_upscaler/watch.py - Folder watcher → Processed archive: moves processed
image_upscaler/watch.py - Batch inference → Processing settings: reads defaults
image_upscaler/inference.py - Batch inference → Application logging: configures logging
image_upscaler/inference.py - Batch inference → Directory and timing helpers: creates directories
image_upscaler/inference.py - Batch inference → Image discovery: scans and loads
image_upscaler/inference.py - Image discovery → Processing settings: reads formats
image_upscaler/src/image_loader.py - Batch inference → Model loading: loads selected models
image_upscaler/inference.py - Model loading → Processing settings: resolves model names
image_upscaler/src/model_loader.py - Model loading → Model weights: reads checkpoints
image_upscaler/src/model_loader.py - Batch inference → Single upscaler: runs single model
image_upscaler/inference.py - Batch inference → Ensemble upscaler: runs ensemble
image_upscaler/inference.py - Single upscaler → Directory and timing helpers: times inference
image_upscaler/src/upscaler.py - Ensemble upscaler → Directory and timing helpers: times inference
image_upscaler/src/upscaler.py - Batch inference → Upscaled images: saves results
image_upscaler/inference.py - User → Channel verification: runs diagnostics
README.md - Channel verification → Model weights: loads checkpoint
verify_channels.py - Channel verification → Comparison images: writes comparisons
verify_channels.py - PSNR comparison → Comparison images: reads comparisons
check_psnr.py - Sharpness analysis → Comparison images: reads comparisons
measure_sharpness.py - User → PSNR comparison: runs diagnostics
README.md - User → Sharpness analysis: runs diagnostics
README.md