---
title: Peripheral Blood Cell Analysis
emoji: 🔬
colorFrom: blue
colorTo: red
sdk: gradio
sdk_version: 4.19.2
app_file: app.py
pinned: false
license: mit
---
# Peripheral Blood Cell Classification with Vision Language Models

## Overview
This application uses state-of-the-art Vision Language Models to classify and describe peripheral blood cells. The system combines advanced computer vision with natural language processing to provide detailed analysis of blood cell images, making it a valuable tool for hematological research and diagnosis.

## Models
The application integrates three powerful Vision Language Models:

1. **Blood Cell Classifier with Llama-3.2**
   - Based on Llama architecture
   - Fine-tuned specifically for blood cell classification
   - Model: laurru01/Llama-3.2-11B-Vision-Instruct-ft-PeripherallBloodCells

2. **Blood Cell Classifier with Qwen2-VL**
   - Leverages Qwen2's vision-language capabilities
   - Optimized for medical image analysis
   - Model: laurru01/Qwen2-VL-2B-Instruct-ft-bloodcells-big

3. **Blood Cell Classifier with SmolVLM**
   - Lightweight yet powerful vision-language model
   - Efficient processing with maintained accuracy
   - Model: laurru01/SmolVLM-Instruct-ft-PeripherallBloodCells

## Features
- Multi-model analysis for comparative results
- Detailed cell type classification
- Comprehensive morphological descriptions
- Support for common image formats
- Real-time processing and analysis
- User-friendly interface

## Cell Types Detected
- Neutrophils
- Lymphocytes
- Monocytes
- Eosinophils
- Basophils

## Technical Details
- Built with Gradio for the interface
- Powered by PyTorch and Transformers
- Optimized for GPU processing
- Uses 4-bit quantization for efficient model loading
- Implements advanced memory management techniques

## Usage
1. Select a model from the dropdown menu
2. Upload an image of a blood cell
3. Wait for the analysis to complete
4. Review the classification and description

## Performance Notes
- Processing time varies by model
- GPU recommended for optimal performance
- Image quality affects accuracy

## Limitations
- Processes one cell at a time
- Requires clear, focused images
- May have varying response times based on server load

## Future Improvements
- Support for batch processing
- Additional model integrations
- Enhanced visualization options
- Performance optimizations

## Citations
If you use this application in your research, please cite:
@software{blood_cell_classifier,
author = {Laurru},
title = {Peripheral Blood Cell Classification with Vision Language Models},
year = {2024},
publisher = {Hugging Face},
url = {https://huggingface.co/spaces/laurru/blood-cell-classifier}
}
## Contact
For questions or suggestions, please open an issue in the repository or contact through Hugging Face.

## License
This project is licensed under the MIT License - see the LICENSE file for details.