ArcheoVLM

Comprehensive Project Execution Checklist

Complete task breakdown for all project phases

Phase 0

Phase 1: Cloud Foundation & Data Ingestion

Create new Google Cloud Project

Write and test setup_gcp.sh script to create GCS bucket and folder structure

Execute setup_gcp.sh script

Verify GCS bucket and folder structure are correct

Develop ingest_data.py script with parallelization, checkpointing, and logging

Test ingest_data.py on a small subset of LiDAR files

Execute full data ingestion process

Verify all raw LiDAR files are successfully stored in /00_raw_lidar/

Phase 1

Phase 2: Automated Triage & Prioritization

Acquire and pre-process all necessary GIS data layers

Store GIS data in /01_inventory_and_gis_data/

Develop triage_tiles.py script for automated spatial cross-referencing

Implement automated prioritization algorithms

Run final triage_tiles.py script to categorize all tiles

Verify that raw.laz files are correctly copied to /02_filtered_tiles/ subdirectories

Phase 2

Phase 3: Scalable LiDAR Processing

Set up GKE Autopilot cluster environment

Containerize the process_lidar_tile.py script and dependencies

Develop and fine-tune the Point Transformer V3 model

Develop and parameter-tune the Cloth Simulation Filter (CSF)

Implement the adaptive interpolation logic in the script

Implement the rvt-py visualization generation logic

Test the full process_lidar_tile.py job on a single high-potential tile

Execute the batch processing job on all prioritized tiles in GKE Autopilot

Verify that all output COGs are correctly generated

Phase 3

Phase 4: Hybrid Intelligence Detection

Set up the annotation environment (e.g., Labelbox, CVAT)

Have expert archaeologist annotate the initial seed set of ~200 RVT images

Configure and train the initial YOLOv9 model on the seed set

Develop the VLM analysis script with CoT prompting

Develop the Active Learning module script

Begin Active Learning Loop iterations

Run YOLO and VLM inference on the unlabeled pool

Consolidate final detections from both YOLO and VLM

Phase 4

Phase 5: Georeferencing & Verification

Develop generate_verification_package.py script

Implement georeferencing function to convert pixel coordinates

Set up API access for Google Earth Engine (Sentinel-2)

Set up access to L-band SAR data archives

Digitize and create the historical text corpus

Implement the GeospaCy/GeoNorm toponym resolution pipeline

Identify and query historical aerial photography archives

Run the generate_verification_package.py script for all sites

Verify all verification packages are complete

Phase 5

Phase 6: Collaborative Review & Dissemination

Establish the formal collaborative review committee

Schedule and conduct review meetings to evaluate verification packages

Create a final, validated list of newly discovered sites

Develop the interactive dashboard template

Generate a unique, secure dashboard for each validated site

Draft scientific publications with partners as co-authors

Submit publications for peer review

Distribute project reports and data to partners

Archive project data and code according to the governance plan

Project Completion Metrics

Key indicators of successful project execution

Data Processing

~3,154 LiDAR tiles processed

AI Performance

Optimized detection accuracy

Archaeological Discovery

Verified site identifications