cq-ai / hive-net

HIVE-Net: Centerline-Aware HIerarchical View-Ensemble Convolutional Network for Mitochondria Segmentation in EM Images

HIVE-Net-Centerline-Aware-HIerarchical-View-Ensemble-Convolutional-Network-for-Mitochondria-Segment

Here are implementations for paper:

HIVE-Net: Centerline-Aware HIerarchical View-Ensemble Convolutional Network for Mitochondria Segmentation in EM Images.

Contact: Zhimin Yuan ([email protected])

Network Structure

Requirements

• Tested with Python 3.6
• CUDA 9.0 or higher
• PyTorch 1.1.0
• numpy 1.16.4
• albumentations 0.3.0

Task

Mitochondria segmentation

Dataset

1. EPFL dataset: https://www.epfl.ch/labs/cvlab/data/data-em/
2. Kasthuri++ dataset: https://sites.google.com/view/connectomics/home

The data augmentation library Albumentation

Training

Run main.py either in your favorite Python IDE or the terminal by typing:

python main.py

Testing

Run mito_prediction.py either in your favorite Python IDE or the terminal by typing:

python mito_prediction.py