README.md

metadata

title: Soil profile analysis
emoji: 🏗️
colorFrom: blue
colorTo: green
sdk: streamlit
sdk_version: 1.28.0
app_file: app.py
pinned: false
license: mit

🏗️ Soil Boring Log Analyzer

An AI-powered application for analyzing soil boring logs using multiple LLM providers. Upload PDF or image files of soil boring logs to automatically extract and analyze soil layers with professional geotechnical insights.

✨ Features

• Multi-LLM Support: Choose from OpenRouter, Anthropic Claude, or Google Gemini
• Document Processing: Upload PDF or image files of soil boring logs
• AI Analysis: Three analysis methods (CrewAI, LangGraph, Unified Workflow)
• Soil Classification: Automatic soil type classification with strength parameters
• Interactive Visualizations: Professional soil profile charts with units
• Layer Processing: Smart layer merging and splitting capabilities

🚀 Quick Start

1. Configure LLM Provider: Add your API key for at least one provider:

   • OpenRouter: Get key from openrouter.ai/keys
   • Anthropic: Get key from console.anthropic.com
   • Google AI Studio: Get key from aistudio.google.com/app/apikey

2. Upload Document: Choose a soil boring log file (PDF, PNG, JPG)

3. Select Analysis Method:

   • CrewAI: Two-agent system with quality control
   • LangGraph: Single agent workflow
   • Unified Workflow: Streamlined processing

4. Configure Options: Set layer merging and splitting preferences

5. Analyze: Get detailed soil analysis with interactive visualizations

🔧 Supported File Formats

• PDF: Soil boring log reports
• Images: PNG, JPG, JPEG of soil boring logs

🤖 LLM Providers

OpenRouter

• Access to multiple models through one API
• Recommended: Claude-4.0 Sonnet, GPT-4 Turbo

Anthropic Direct

• Direct access to Claude models
• Excellent for technical analysis

Google AI Studio

• Direct access to Gemini models
• Advanced multimodal capabilities

📊 Analysis Methods

CrewAI (Recommended)

• Soil Expert Agent: Specializes in soil classification
• Geotechnical Agent: Focuses on engineering parameters
• Quality Control: Two-agent validation system

LangGraph

• Single agent with structured workflow
• Good for straightforward analyses

Unified Workflow

• Streamlined processing pipeline
• Fast analysis with comprehensive validation

🔬 Technical Features

• Su Detection: Comprehensive undrained shear strength value extraction
• Layer Optimization: Smart merging of similar layers
• Thick Layer Splitting: Automatic subdivision of thick layers
• Unit Conversions: Proper handling of different measurement units
• Professional Charts: Publication-ready soil profile visualizations

🎯 Use Cases

• Geotechnical Engineering: Foundation design analysis
• Site Investigation: Soil characterization studies
• Construction Planning: Subsurface condition assessment
• Academic Research: Soil mechanics studies
• Consulting: Client report generation

🔒 Privacy & Security

• No Data Storage: Documents are processed in memory only
• API Key Security: Keys are handled securely in environment variables
• Local Processing: All analysis happens in your session

🛠️ Technical Stack

• Frontend: Streamlit
• AI Framework: LangGraph, CrewAI
• LLM Integration: OpenAI API, Anthropic API, Google AI
• Visualization: Plotly, Matplotlib
• Document Processing: PyPDF2, PIL

📝 License

This project is developed for geotechnical engineering applications with AI-powered analysis capabilities.