A Streamlit application for running YOLOv11 inference and reviewing/correcting annotations.
- Load YOLOv11 models from weights folder or upload custom models
- Configure inference parameters (confidence threshold, IoU, image size, etc.)
- Process images or videos from a directory or upload files
- Export results in multiple formats (TXT, CSV)
- Save cropped detections for easy review
- Review annotations in manageable chunks
- Visualize all detections in an image
- Review individual cropped detections
- Reclassify detections with a simple dropdown interface
- Track review progress
- Export finalized annotations in CSV and YOLO formats
- Clone this repository
- Install the required packages:
pip install -r requirements.txt
- Run the Streamlit application:
streamlit run app.py
-
Open your browser and navigate to the URL displayed in the terminal (usually http://localhost:8501)
-
Use the navigation sidebar to switch between pages:
- Run Inference: Execute YOLOv11 model on selected data
- Annotation Review: Review and correct model predictions
- Select or upload a YOLOv11 model
- Choose your data source (upload files or specify a directory)
- Configure model parameters
- Set output options
- Click "Run Inference" to start processing
- Download the results when complete
- Enter the path to the review folder (created during inference)
- Set how many images you want to review in the current session
- Review and correct class predictions as needed
- Mark images as reviewed
- Save your changes
- Export the final annotations in your preferred format (CSV or YOLO)
app.py: Main application entry pointpages/: Contains the individual application pagesrun_inference.py: Code for the inference pageannotation_review.py: Code for the annotation review page
weights/: Directory for storing model weightsoutput/: Directory where results are saved