Update README.md

README.md

@@ -1,195 +1,11 @@
+---
+title: Fungi Classifier
+emoji: 🍄
+colorFrom: green
+sdk: streamlit
+app_file: app.py
+---
 
-# Dependable FungAI - README
-
-## Overview
-
-This project "Dependable FungAI" implements an **Enhanced Hybrid Concept-Based Mushroom Classification System** using advanced machine learning techniques including **CLIP (ViT-L/14)** for visual feature extraction and **Multi-Layer Perceptron (MLP)** networks. The system classifies mushrooms as either **edible** or **poisonous** based on both visual features and textual concept descriptions. The project generates comprehensive **Training Bill of Materials (TBOM)** and **Inference Bill of Materials (IBOM)** for complete transparency, traceability, and explainability.
-
-## Key Features
-
-- **Enhanced IBOM Analysis**: Interactive inference with concept categorization by morphological features
-- **TBOM Visualization**: Comprehensive training documentation and performance metrics
-- **Interactive Concept Modification**: Real-time adjustment of concept scores with impact visualization
-- **Educational Mode**: Detailed explanations, safety guidance, and identification tips
-- **DSSE Signature Support**: Supply chain security with in-toto attestation
-- **Export Functionality**: Multiple download formats (JSON, CSV, comprehensive reports)
-- **Advanced Safety Features**: Conflict detection and biological impossibility validation
-
-## System Architecture
-
-### Training Components (TBOM)
-- **Enhanced Hybrid MLP Classifier** with concept-based learning
-- **CLIP ViT-L/14** backbone for visual feature extraction
-- **117 mushroom-specific concepts** generated from morphological features
-- **Cross-validation** with comprehensive performance tracking
-
-### Inference Components (IBOM)
-- **Interactive concept categorization** by morphological features:
-  - Cap properties (color, shape, surface)
-  - Gill properties (color, spacing, attachment)
-  - Stalk properties (shape, color, texture)
-  - Sensory properties (odor detection)
-  - Environmental context (habitat, growth patterns)
-  - Reproductive features (spore characteristics)
-
-## Key Terms Defined
-
-- **TBOM (Training Bill of Materials)**: Comprehensive documentation of the training process, model architecture, performance metrics, and validation results
-- **IBOM (Inference Bill of Materials)**: Detailed analysis of individual predictions including concept contributions, uncertainty visualization, and safety assessments
-- **DSSE (Dead Simple Signing Envelope)**: Cryptographic signing standard for supply chain security and integrity verification
-- **Concept Categorization**: Organization of mushroom identification features by morphological categories for educational clarity
-- **Uncertainty Visualization**: Interactive charts showing prediction confidence, decision boundaries, and risk assessment
-- **Conflict Detection**: Automated identification of biologically impossible feature combinations
-
-## Requirements
-
-### Data Set
-
-1. For **CSV Dataset**: https://www.kaggle.com/datasets/uciml/mushroom-classification
-2. For **Image Classification:** https://www.kaggle.com/datasets/maysee/mushrooms-classification-common-genuss-images
-3. For **Multi- Modal Approach** (Images dataset): 
-git clone: https://www.kaggle.com/datasets/derekkunowilliams/mushrooms
-
-This has four folders: Edible, Conditionally Edible, Poisonous, and Deadly Poisonous. Merge Edible and Conditionally Edible into one Edible folder, and Poisonous and Deadly Poisonous into one Poisonous folder, for a binary classification setup.
-
-### Core Dependencies
-
-pip install streamlit torch clip-by-openai pillow plotly pandas numpy scikit-learn opencv-python
-
-
-### System Requirements
-- Python 3.8+
-- CUDA-compatible GPU (optional, falls back to CPU)
-- 8GB+ RAM recommended for CLIP model loading
-
-## Installation & Setup
-
-1. **Clone the repository**:
-
-git clone https://github.com/umaima786/dependable-fungai.git
-cd dependable-fungai
-
-
-2. **Install dependencies**:
-
-pip install -r requirements.txt
-
-
-3. **Set up environment variables** (optional but recommended):
-
-export TBOM_PATH="/path/to/TBOM.json"
-export MODEL_PATH="/path/to/final_model.pt"
-export CSV_PATH="/path/to/mushrooms.csv"
-
-
-## Usage Instructions
-
-### 1. Training the Model (TBOM Generation)
-
-**Run the training script**:
-
-python TBOM.py
-
-
-**Interactive prompts will ask for**:
-- Path to mushroom CSV dataset (or press Enter for default)
-- Path to mushroom image dataset (or press Enter for default)
-
-**Outputs**:
-- `TBOM.json`: Complete training documentation
-- `final_model.pt`: Trained model weights
-- Comprehensive performance metrics and validation results
-
-### 2. Running the Interactive Dashboard
-
-**Start the Streamlit application**:
-
-streamlit run app.py
-
-
-**Features available**:
-- **TBOM Analysis Tabs**:
-  - Data Summary: Dataset statistics and class distribution
-  - Overview: Performance metrics and comparison charts
-  - Performance: Confusion matrices, ROC/PR curves, training progress
-  - Architecture: Model structure and component visualization
-  - Concepts: Concept analysis and model interpretation
-  - Technical: Environment details and raw TBOM data
-
-- **Enhanced IBOM Tab**:
-  - Interactive image upload and analysis
-  - Real-time concept score modification by category
-  - Uncertainty visualization with confidence intervals
-  - Educational mode with safety guidance
-  - Export functionality for analysis results
-
-### 3. Running Inference with IBOM Generation
-
-**Command-line inference**:
-
-python IBOM.py --image_file mushroom.jpg --dsse --educational_mode
-
-
-**Batch processing**:
-
-python IBOM.py --image_dir images/ --output analysis.json --export_summary
-
-
-**Parameters**:
-- `--image_file`: Single mushroom image for analysis
-- `--image_dir`: Directory containing multiple images
-- `--output`: Custom output filename
-- `--dsse`: Include DSSE signature for security
-- `--educational_mode`: Enable detailed explanations
-- `--export_summary`: Generate additional summary report
-
-## Safety & Educational Features
-
-### Critical Safety Notices
-- **NEVER consume any mushroom based solely on AI analysis**
-- Always consult certified mycologists for positive identification
-- Misidentification can be fatal
-
-### Educational Enhancements
-- **Morphological categorization** with biological context
-- **Interactive learning** through concept modification
-- **Conflict detection** for impossible feature combinations
-- **Uncertainty quantification** with confidence interpretation
-- **Safety guidance** with risk assessment
-
-## Advanced Features
-
-### Security & Traceability
-- **DSSE signatures** for supply chain integrity
-- **Cryptographic hashing** of analysis results
-- **Complete audit trail** from training to inference
-
-### Export & Integration
-- **JSON exports** for programmatic access
-- **CSV data** for spreadsheet analysis
-- **Comprehensive reports** for documentation
-- **API-ready format** for system integration
-
-## Troubleshooting
-
-### Common Issues
-
-1. **Model files not found**:
-   - Run `python TBOM.py` first to generate required files
-   - Set environment variables for custom paths
-
-2. **Memory errors**:
-   - Reduce batch size in training
-   - Use CPU instead of GPU if VRAM insufficient
-
-3. **Interactive prompts in production**:
-   - Set all environment variables to avoid prompts
-   - Use `--no_interactive` flag for automated deployment
-
-### Getting Help
-
-- Check the **Technical Details** tab in the dashboard for system information
-- Enable **detailed error display** in the IBOM interface for debugging
-- Consult the **Educational Mode** for feature explanations
-
+# Fungi Classifier Application
 
+This is a Streamlit application for classifying mushrooms.