wumingyu123/bccd-blood-cell-detection

BCCD - Blood Cell Count and Detection Dataset

Dataset Description

BCCD (Blood Cell Count and Detection) is a dataset for blood cell object detection. It contains microscopic blood smear images with annotations for three types of blood cells.

Classes

Class	Description
RBC	Red Blood Cell (Erythrocyte)
WBC	White Blood Cell (Leukocyte)
Platelets	Thrombocyte

Dataset Structure

BCCD/
├── Annotations/          # XML annotation files (Pascal VOC format)
├── JPEGImages/           # Blood cell images (640x480)
└── ImageSets/
    └── Main/
        ├── train.txt     # Training set (205 images)
        ├── val.txt       # Validation set (87 images)
        ├── test.txt      # Test set (72 images)
        └── trainval.txt  # Train+Val (292 images)

Statistics

Split	Images	Annotations
Train	205	~2,805
Val	87	~1,138
Test	72	~945
Total	364	~4,888

Citation

If you use this dataset, please cite the original BCCD dataset.

Source

Original: BCCD_Dataset