Back to challenges
Challenge

Build a Multimodal Building Inspection Agent with Google ADK and Gemini

Engineering teams manually inspecting legacy office blueprints miss critical MEP and core structural bottlenecks prior to conversion planning. Build a Google ADK multimodal agent leveraging Gemini to process architectural floor plans and output structural feasibility reports with 88% recall on risk flags.

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 AI agent using Google Agent Development Kit (ADK) and Gemini to analyze architectural blueprint images and structural text records.

Delivery guide

How work is evaluated

Evaluation

Evaluates detection precision and visual recall for architectural blueprint structural bottleneck analysis.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Loading datasets...
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

Risk Recall Test

Ensures all known architectural blockers in test images are flagged.

Binary check

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

Dimension 2

Visual Inspection Recall

Percentage of actual structural bottlenecks recalled. • target: 0.88 • 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

  • Implement multimodal image and text processing pipelines using Google ADK and Gemini.

  • Extract structural elements (shear walls, elevator cores, plumbing stacks) from blueprint images.

  • Generate standardized conversion risk matrices linking visual findings to engineering constraints.

Resources and assets

Reference links and supporting material

Dataset notes

Multimodal dataset containing office blueprint layout images, spatial metadata, and expert structural audit reports.

How this agent runs

Evaluates detection precision and visual recall for architectural blueprint structural bottleneck analysis.

Challenge input

JSON with image_uri, floor_level, structure_type

Google ADK

Assigned framework for building Vertex AI and Gemini agents.

Evaluated output

JSON with load_bearing_wall_detected, core_shaft_count, plumbing_riser_bottlenecks, risk_flag_count

Checks for
  • Ensures all known architectural blockers in test images are flagged.
  • Percentage of actual structural bottlenecks recalled. • target: 0.88 • range: 0-1
Proof of success
  • Benchmark: Multimodal_Architecture_Eval
  • Visual Inspection Recall target: 0.88
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

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

Evaluation contract

  • Risk Recall Test · Weight 1
  • Visual Inspection Recall · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about Build a Multimodal Building Inspection Agent with Google ADK and Gemini