AI FEATURES

Photo estimation

Upload a site photograph. ZRO identifies construction elements, estimates visible quantities, and produces a structured preliminary estimate mapped to CWICR positions — in a single local inference pass.

8 min read

On this page

How it works Capturing effective photographs Understanding the output Limitations

How it works

The photo is passed to a local multimodal model that analyses the image without sending any data off your server. The model performs three operations in sequence:

Element identification

The model scans the image and identifies visible construction elements — structural members, enclosure systems, internal finishes, MEP runs, site works, and site conditions. Each element is described with type, estimated material, and a qualitative assessment of quantity visibility.

Quantity estimation

For each identified element, the model estimates visible quantities using proportional analysis — comparing element dimensions against reference objects in the frame (doors, windows, figures, standard brickwork courses). Where no scale reference is present, the model applies construction-typical dimensions and flags the quantity as low-confidence.

CWICR mapping

Each element is matched against the CWICR database for the project's regional market. The closest matching position is selected based on material, construction method, and application. The result is a structured line-item list ready for import into a BOQ.

Capturing effective photographs

Practice	Why it matters
Include a scale reference	A person, door, or surveyor's staff in the frame dramatically improves quantity confidence
Multiple angles	Use 3–5 photos for complex areas rather than one wide shot
Avoid extreme wide-angle	Lens distortion makes dimensional inference unreliable
Good lighting, no harsh shadows	Shadows obscure material identification
Resolution: 2 MP minimum	Higher resolution improves small-element identification
Date-stamp not required	The system timestamps on upload

Tip

For structural photography, photograph from approximately 10 metres at eye level. For finishes and cladding, include at least 3 × 3 metres of continuous surface in frame.

Understanding the output

Each photo estimate returns:

json

{
  "summary": "Ground-floor reinforced concrete frame visible. External masonry walls under construction. Timber formwork in place for first-floor slab.",

  "elements": [
    {
      "description": "Reinforced concrete columns, approximately 350×350mm",
      "quantityConfidence": "medium",
      "note": "Scale inferred from adjacent doorway — verify on drawings"
    }
  ],

  "lineItems": [
    {
      "cwicrCode": "02.01.010",
      "description": "Reinforced concrete column, 350×350mm, C30/37",
      "quantity": 8,
      "unit": "no.",
      "unitRate": 1840.00,
      "total": 14720.00,
      "confidence": 0.74
    },
    {
      "cwicrCode": "02.01.020",
      "description": "Reinforced concrete suspended slab, 200mm, C25/30",
      "quantity": 420,
      "unit": "m²",
      "unitRate": 113.00,
      "total": 47460.00,
      "confidence": 0.82
    }
  ],

  "subtotal": 186340,
  "contingency": 18634,
  "total": 204974,
  "currency": "EUR",
  "overallConfidence": 0.77
}

Confidence scores:

Score	Interpretation	Recommended action
0.90 – 1.00	High — element clearly visible with reliable scale reference	Use directly in BOQ with standard review
0.70 – 0.89	Medium — element identified but quantity uncertain	Verify quantities against drawings
0.50 – 0.69	Low — element inferred or partially visible	Treat as order-of-magnitude only
< 0.50	Very low — uncertain identification	Do not carry forward without manual verification

Limitations

Important

Photo estimation is a feasibility and preliminary design tool. It cannot substitute a measured survey or full quantity takeoff for tender purposes.

Limitation	Detail
Hidden works	Foundations, drainage, and embedded services are not visible in photographs
Substructure	Below-ground elements require separate measurement
Finishes detail	Paint, sealants, ironmongery, and specialist finishes are frequently missed
MEP systems	Ductwork and cabling can be identified in exposed areas but not routed elements
Complex geometry	Non-rectilinear forms introduce dimensional uncertainty
Night photography	Low-light images significantly reduce accuracy

← PreviousCopilot modesAI FEATURES Next →Document intelligenceAI FEATURES