Skip to main content
Back to challenges
Challenge

Build Multimodal Zoning Document & Site Plan Inspector using Google ADK

Verifying large-scale urban regeneration master plans against municipal zoning frameworks requires parsing multimodal inputs (blueprints, aerial imagery, site boundary vector data, and zoning PDF texts). Developers will construct a multimodal site inspector agent using Google ADK. The agent uses Gemini's vision and document capabilities to analyze urban site maps alongside zoning bylaws, automatically identifying unauthorized land use proposals and FAR violations.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Construct a multimodal zoning document inspection agent using Google ADK and Gemini.

Delivery guide

How work is evaluated

Evaluation

Evaluates detection accuracy of unauthorized land uses and FAR calculation precision.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Land Use Compliance Check

Ensures unpermitted land use triggers compliance failure

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

FAR Calculation Precision

Accuracy of computed Floor Area Ratio • target: 1 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Utilize Google ADK for multimodal document processing of site plans and zoning text

  • Implement vision-based boundary verification algorithms using Gemini 1.5

  • Flag non-compliant land use codes in mixed-use development proposals

  • Generate automated compliance certificates for planning commissions

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of 30 urban site plans and municipal zoning code rulebooks.

Challenge setup

Evaluates detection accuracy of unauthorized land uses and FAR calculation precision.

Challenge input

JSON with district_zone, proposed_use, building_sqft, lot_sqft, max_far

Google ADK

Required framework for Gemini agent.

Evaluated output

JSON with compliant (boolean), calculated_far, land_use_allowed (boolean), violation_details

Evaluation checks
  • Ensures unpermitted land use triggers compliance failure
  • Accuracy of computed Floor Area Ratio • target: 1 • range: 0-1
Configured evaluation evidence
  • Benchmark: MultimodalZoningBench
  • FAR Calculation Precision target: 1
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Environment
  • Modal · Optional
Action Space
  • Google ADK · Required
  • Google · Optional
Policy Serving
  • Modal · Optional

Evaluation contract

  • Land Use Compliance Check · Weight 1
  • FAR Calculation Precision · Weight 1

Metrics

  • FAR Calculation Precision · Target 1

Benchmarks

  • MultimodalZoningBench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with district_zone, proposed_use, building_sqft, lot_sqft, max_far

    Output: JSON with compliant (boolean), calculated_far, land_use_allowed (boolean), violation_details

    Execution harness: Python

Evaluation tools

  • Google ADK · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build Multimodal Zoning Document & Site Plan Inspector using Google ADK