https://github.com/deniztemur00/vindrmammo
Deep learning project for automated detection of breast abnormalities in mammogram images
https://github.com/deniztemur00/vindrmammo
computer-vision deep-learning detection feature-engineering
Last synced: 9 months ago
JSON representation
Deep learning project for automated detection of breast abnormalities in mammogram images
- Host: GitHub
- URL: https://github.com/deniztemur00/vindrmammo
- Owner: deniztemur00
- Created: 2025-01-05T12:40:54.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-09-23T20:37:31.000Z (10 months ago)
- Last Synced: 2025-09-23T22:22:21.861Z (10 months ago)
- Topics: computer-vision, deep-learning, detection, feature-engineering
- Language: Python
- Homepage:
- Size: 4.9 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# VinDr-Mammo Object Detection
Deep learning project for automated detection of breast abnormalities in mammogram images using the VinDr-Mammo dataset.
## Key Features
- Faster R-CNN and RetinaNet architecture for multi-class object detection
- Balanced dataset creation through stratified sampling based on:
- BI-RADS classifications (1-5)
- Breast density categories (A-D)
- Finding categories (Mass, Calcification, Asymmetry, etc.)
- Data pruning to address class imbalance:
- Reduced overrepresented "No Finding" class from 18k to 2.5k samples
- Preserved rare finding combinations across train/test splits
## Setup
### Clone the repository
```bash
git clone https://github.com/yourusername/vindr-mammo.git
```
### Install dependencies
```bash
pip install -r requirements.txt
```
## Results
TODO: Add results as a notebook