Repository navigation
Releases
Releases · AhmadJeddi/Resnet-DCBAM
ResNet-DCBAM v1.0.1 - Enhanced Testing Example
What's Changed
This patch release improves the backbone testing example.
Improvements
- Updated
examples/test_backbone.pyto demonstrate layer freezing functionality. - Added configurable
freeze_layerssupport in the example. - Updated input size and batch handling for a more realistic forward pass test.
- Added feature shape reporting for backbone outputs.
- Added parameter statistics:
- Total parameters
- Frozen parameters
- Trainable parameters
- DCBAM-specific parameter statistics
Notes
This release does not change the model architecture or public API.
It only improves the example workflow and provides better debugging and verification utilities.
Full Changelog:
v1.0.0 → v1.0.1
ResNet-DCBAM v1.0.0 - Initial Release
Initial Release
First public release of ResNet-DCBAM.
This release provides a modular PyTorch implementation of a ResNet50 backbone enhanced with Dual Channel and Spatial Attention mechanisms.
Features
- Modular DCBAM attention implementation
- ResNet50 backbone with optional attention integration
- Layer freezing and pretrained weight support
- Example usage and testing scripts
- Clean project structure with documentation
Technologies
- Python
- PyTorch
- Deep Learning
- Computer Vision