ArchaeoVLM: Archaeological Vision Language Model

Multi-pipeline system for archaeological remote sensing analysis with cost-effective cloud processing.

🎯 Overview

ArchaeoVLM provides specialized pipelines for archaeological analysis:

• LiDAR Processing: RunPod GPU + Google Cloud Storage (70% cost savings)
• Data Ingestion: Multi-source archaeological data integration
• Model Training: Custom archaeological feature detection
• Results Analysis: Visualization and reporting tools

📁 Project Structure

🔄 LiDAR Visualization System

• lidar_visualization/ - Complete LiDAR processing and analysis system
  • pipelines/lidar_processing/ - LiDAR semantic segmentation with PTv3
    • runpod_gcs_archaeovlm.py - Main processing engine
    • setup_runpod_gcs.sh - One-click environment setup
    • example_usage.py - Usage examples and scenarios
  • pipelines/data_ingestion/ - Multi-source data integration
  • pipelines/model_training/ - Custom model training workflows
  • pipelines/results_analysis/ - Visualization and analysis tools
  • shared/ - Shared components across pipelines
    • models/ - Point Transformer V3 and other models
    • utils/ - Common utilities and helpers
    • configs/ - Shared configuration templates

📚 Documentation & Resources

• docs/ - Complete project documentation
  • index.html - Main project visualization interface
  • lidar_visualization/ - LiDAR system documentation
    • RUNPOD_GCS_GUIDE.md - Quick start guide (30 min setup)
    • GCP_RUNPOD_DEPLOYMENT_GUIDE.md - Comprehensive deployment guide
• data/ - Local data directory

🚀 Quick Start (LiDAR Processing Pipeline)

# 1. Setup Google Cloud Storage
gcloud projects create archaeovlm-project
gsutil mb gs://archaeovlm-lidar-data
gsutil mb gs://archaeovlm-results

# 2. Launch RunPod RTX 3090 instance ($0.34/hr)

# 3. Setup environment
cd lidar_visualization/pipelines/lidar_processing/
wget https://raw.githubusercontent.com/your-repo/lidar_visualization/pipelines/lidar_processing/setup_runpod_gcs.sh
chmod +x setup_runpod_gcs.sh
./setup_runpod_gcs.sh

# 4. Upload GCP credentials
scp gcp-key.json root@[runpod-ip]:/workspace/gcp-credentials.json

# 5. Start processing
python3 runpod_gcs_archaeovlm.py gs://archaeovlm-lidar-data/

💰 Cost Examples

Scenario	Files	Budget	Use Case
Pilot Study	10 files	$2-3	Quick validation
Site Survey	25 files	$5-8	Single site analysis
Regional Study	50 files	$10-15	Multi-site comparison
Research Project	100 files	$20-30	Publication dataset

🔧 Configuration (LiDAR Pipeline)

Edit lidar_visualization/pipelines/lidar_processing/archaeovlm_runpod_gcs_config.json:

{
  "processing": {
    "max_files": 25,
    "preferred_regions": ["TAP", "FN2", "ANT"],
    "file_selection": "size_based",
    "date_range": ["2010", "2020"]
  },
  "storage": {
    "project_id": "your-gcp-project",
    "bucket": "your-results-bucket",
    "input_bucket": "your-lidar-data"
  }
}

✨ Features

• Smart file selection by archaeological region and date
• Automatic cost estimation before processing
• Cloud backup of all results
• Auto-shutdown to prevent idle costs
• Real-time monitoring with cost tracking
• Compression to reduce storage costs by 70%

📊 Output

• Classified point clouds with archaeological labels
• Processing reports with statistics
• Cost tracking and quality metrics
• Ready for QGIS analysis and publication

🛠️ Pipeline Support

LiDAR Processing Pipeline

cd lidar_visualization/pipelines/lidar_processing/

# Test setup
python3 test_runpod_gcs_setup.py

# Monitor processing
python3 monitor_runpod_gcs.py

# Check logs
tail -f /workspace/logs/archaeovlm.log

📚 Documentation

LiDAR Processing

• Quick Start: docs/lidar_visualization/RUNPOD_GCS_GUIDE.md (30 min setup)
• Complete Guide: docs/lidar_visualization/GCP_RUNPOD_DEPLOYMENT_GUIDE.md (detailed)
• Examples: lidar_visualization/pipelines/lidar_processing/example_usage.py (different scenarios)
• Web Interface: docs/index.html (main project dashboard)

Other Pipelines

• Data Ingestion: Coming soon
• Model Training: Coming soon
• Results Analysis: Coming soon

---

Current Focus: LiDAR Processing Pipeline
Platform: RunPod GPU + Google Cloud Storage
Cost: $0.34/hr + $0.02/GB storage
Perfect for: Budget-conscious archaeological research