Skip to content

SENLA-ResUNet v1.0.0 – Pretrained Nuclei Segmentation Model

Latest
Latest

Choose a tag to compare

View all tags
@OVER-CODER OVER-CODER released this 18 Aug 14:44
· 4 commits to main since this release
v1.0.0
6442be4
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

SENLA-ResUNet v1.0 – Pretrained Nuclei Segmentation Models

📢 First official release of SENLA-ResUNet, our hybrid Residual U-Net architecture with Squeeze-and-Excitation and Non-Local Attention for robust nuclei segmentation in biomedical images.

🔹 What’s Included

  • Pretrained model weights:
    • SENLA_ResUnet_tuned_trained.h5
    • SENLA_ResUnet_tuned_trained.keras
  • Training scripts and inference utilities
  • Dataset preprocessing instructions
  • Evaluation metrics and example predictions

🔹 Performance Highlights

  • DSB2018

    • IoU = 0.7936
    • Dice = 0.8172
    • Precision = 0.8791
    • Recall = 0.9084
    • Accuracy = 0.9726

  • TNBC

    • IoU = 0.8332
    • Dice = 0.8986
    • Precision = 0.9353
    • Recall = 0.8822
    • Accuracy = 0.9582

🔹 How to Use

Download pretrained weights:

wget https://github.com/OVER-CODER/Nuclei-Segmentation/releases/download/v1.0.0/SENLA_ResUnet_tuned_trained.h5
wget https://github.com/OVER-CODER/Nuclei-Segmentation/releases/download/v1.0.0/SENLA_ResUnet_tuned_trained.h5
Assets 4
Loading