How atifamjad707/breast-cancer-yolov9-detection works
This repository is a command-line research pipeline for detecting Normal and Breast-cancer regions in mammograms. The user downloads the Roboflow dataset, preprocesses its splits with CLAHE, trains a YOLOv9 model, then evaluates its metrics or runs inference on new images. The scripts are separate entry points coordinated by the user and by files passed between runs, rather than one runtime orchestrator. The README documents the stages and results; the sampled source confirms the script behavior and Ultralytics calls.
Main entry points: evaluate.py Model evaluation · predict.py Image prediction · download_dataset.py Dataset downloader · preprocess.py CLAHE preprocessing
Components 14
Dataset acquisition
- Dataset downloader
src/download_dataset.py - YOLO dataset
Data preparation
- CLAHE preprocessing
src/preprocess.py - Processed dataset
Training and evaluation
- YOLO training
src/train.py - Trained weights
- Model evaluation
src/evaluate.py - Metrics and plots
Inference
- Input mammograms
- Image prediction
src/predict.py - Annotated detections
Other components
- Researcher / CLI user
- Roboflow service
- Ultralytics YOLO
Connections 21
- Researcher / CLI user → Dataset downloader: runs
- Dataset downloader → Roboflow service: downloads from
- Dataset downloader → YOLO dataset: writes
- Researcher / CLI user → CLAHE preprocessing: runs
- CLAHE preprocessing → YOLO dataset: reads
- CLAHE preprocessing → Processed dataset: writes
- Researcher / CLI user → YOLO training: runs
- YOLO training → Processed dataset: reads
- YOLO training → Ultralytics YOLO: trains with
- Ultralytics YOLO → Trained weights: saves
- Researcher / CLI user → Model evaluation: runs
- Model evaluation → Trained weights: loads
- Model evaluation → Processed dataset: reads
- Model evaluation → Ultralytics YOLO: validates with
- Ultralytics YOLO → Metrics and plots: produces
- Researcher / CLI user → Image prediction: runs
- Image prediction → Trained weights: loads
- Researcher / CLI user → Input mammograms: provides
- Image prediction → Input mammograms: passes to inference
- Image prediction → Ultralytics YOLO: predicts with
- Ultralytics YOLO → Annotated detections: saves annotations