This is the code for all models mentioned in the paper "", where the HMIL model is an innovative multiple instance learning model proposed in this paper.
HMIL/
├── MIL/ # Multiple Instance Learning Module
│ ├── models/ # MIL Models
│ │ ├── HMIL.py # HMIL model, including 2 stages: HMILIC and HMILCC
│ │ ├── AttentionMIL.py # Attention MIL model
│ │ ├── TransMIL.py # Transformer MIL model
│ │ └── ClassicalMIL.py # Classical MIL model
│ ├── training_MIL.py # MIL training script
│ ├── inference_MIL.py # MIL inference script (image classification, for HMILIC)
│ ├── inference_HMIL.py # HMIL inference script (case classification, for HMILCC)
│ ├── metrics_MIL.py # MIL evaluation metrics
│ └── data_processor_MIL.py # MIL data processing
│
├── segmentation/ # Image Segmentation Module
│ ├── models/ # Segmentation Models
│ │ ├── AttentionUNet.py # Attention U-Net
│ │ ├── DeepLabv3_plus.py # DeepLabv3+
│ │ └── SegFormer/ # SegFormer Series
│ │ ├── segformer.py
│ │ ├── mix_transformer.py
│ │ └── segformer_head.py
│ ├── training.py # Segmentation model training script
│ ├── inference.py # Segmentation model inference script
│ ├── metrics.py # Segmentation evaluation metrics
│ ├── loss_criteria.py # Loss functions
│ ├── data_processor.py # Data processing
│ └── FPSTest.py # FPS performance test
│
├── LICENSE # License file
└── README.md # Project documentation
- Python>=3.10
- CUDA==12.4
- torch==2.6.0+cu124
- torchvision==0.20.0+cu124
- albumentations==2.0.8
- einops==0.8.1
- huggingface-hub==0.34.3
- monai==1.5.0
- numpy==1.26.4
- opencv-python-headless==4.11.0.86
- openpyxl==3.1.5
- pandas==2.3.1
- pillow==11.0.0
- scikit-learn==1.7.1
- scipy==1.15.3
- timm==1.0.19