IDRiD_Challenge_Mask-RCNN

The Challenge in (https://idrid.grand-challenge.org/)

the the data can be download here

The Mask RCNN's application in IDRiD(only for challenge 1)

原始的数据结构

• Original:

  --traiing

   ---IDRiD_01.jpg
   
   ---IDRiD_02.jpg
   ...
  

  --testing ---IDRiD_01.jpg

   ---IDRiD_02.jpg
   ...
  

• Groundtruths:

  --training

    ---Haemorrhages
      
      IDRiD_01.jpg
      
      IDRiD_02.jpg
      ...
    ---HardExudates
    
    ---Microaneurysms
    
    ---OpticDisc
    
    ---SoftExudates
    
  --testing
    ---Haemorrhages
      
      IDRiD_01.jpg
      
      IDRiD_02.jpg
      ...
    ---HardExudates
    
    ---Microaneurysms
    
    ---OpticDisc
    
    ---SoftExudates
  

• Step 1 : Create Dataset

First,create directory data

mkdir data

And create data

python dataset.py

The data will store in data,they are too big,so I do not show that.

• Step 2: Train the Model

python main.py

the checkpoint will be stored in directory checkpoint,you can download the pretrained model in here

• Step 3: Demo Show the Mask and model evaluation