Fashion & Apparel E-commerce

Business Context

The Problem: Customers don’t know their body measurements. And they refuse to measure themselves. When you ask for their chest or waist size, they guess. When they guess, they bracket-buy — sticking you with the reverse logistics bill and killing your margins through shipping, restocking, and deadstock. Or worse: they abandon the cart entirely.

The Solution: The fix isn’t a better size chart. It’s a better input. The only two numbers every customer actually knows are height and weight. DimensionsPot takes those two inputs and generates a complete, 130-point anthropometric profile in under 10ms. You get the chest, natural waist, hips, and inseam measurements required to confidently recommend a single size at checkout.

The Global Edge: Selling cross-border? A “Size M” in Tokyo has different skeletal proportions than a “Size M” in Berlin. DimensionsPot handles the demographic math natively. By defining the input_origin_region (your customer) and target_region (your brand’s fit model), the API instantly translates body ratios between populations in a single request.

Recommended API Configuration

Threshold note: PRIMARY_BOTH BONE dimensions score ~85; FLESH dimensions ~78. Setting confidence_score_threshold: 80 would suppress most circumference dimensions (FLESH) — exactly the ones needed for apparel sizing. Threshold 75 retains everything clothing-relevant.

Tip — Cross-regional sizing: Set input_origin_region to the customer’s location and target_region to the region your size chart was built for. The Universal Translator eliminates the Double Penalty Paradox that affects naive regional scaling.

Example: Asian customer, European brand → input_origin_region: "ASIA_PACIFIC", target_region: "EUROPE".

Tip — Plus-size / curve lines: Set body_build_type: "OVERWEIGHT" for shoppers with BMI > 30 to activate BMI-stratified NHANES morphing. Circumference dimensions adjust accordingly.

Sample Request

curl -X POST "https://dimensionspot-bodysize-engine.p.rapidapi.com/v1/predict" \
  -H "Content-Type: application/json" \
  -H "X-RapidAPI-Key: YOUR_KEY" \
  -H "X-RapidAPI-Host: dimensionspot-bodysize-engine.p.rapidapi.com" \
  -d '{
    "input_data": {
      "input_unit_system": "metric",
      "subject": {
        "gender": "female",
        "exact_age": 28.0,
        "age_category": "ADULT",
        "input_origin_region": "EUROPE"
      },
      "anchors": {
        "body_height": 1680.0,
        "body_mass": 65.0
      }
    },
    "output_settings": {
      "calculation": {
        "calculation_model": "AUTO",
        "target_region": "EUROPE",
        "body_build_type": "CIVILIAN"
      },
      "requested_dimensions": {
        "bundle": "TORSO",
        "specific_dimensions": null
      },
      "output_format": {
        "unit_system": "metric",
        "confidence_score_threshold": 75,
        "include_range_95": true,
        "include_iso_codes": false
      }
    }
  }'

Key Dimensions for Apparel Sizing

inseam_length, thigh_circumference, and calf_circumference are in the LEGS_FEET bundle. To retrieve them alongside torso dimensions in a single call, use bundle: "FULL_BODY" or list all needed keys explicitly in specific_dimensions.

Waist note: waist_circumference_natural (ISO 6.4.11, narrowest torso point) is the apparel standard. waist_circumference_omphalion (navel level) is the clinical/epidemiological standard (NHANES/WHO). Use waist_circumference_natural for clothing size charts.

Size Chart Mapping — Pattern

Map API dimensions against your stored size chart thresholds client-side. No runtime ML required:

# Example: map torso dimensions to garment sizes
# Adjust thresholds per brand/product category

SIZE_CHART_FEMALE = {
    # (chest_min, chest_max, waist_min, waist_max, hip_min, hip_max) → size label
    "XS": (770, 830,  580, 630,  840, 890),
    "S":  (830, 890,  630, 680,  890, 940),
    "M":  (890, 950,  680, 730,  940, 990),
    "L":  (950, 1010, 730, 800,  990, 1050),
    "XL": (1010, 1100, 800, 880, 1050, 1130),
}

def recommend_size(chest_mm, waist_mm, hip_mm, size_chart):
    candidates = []
    for label, (ch_lo, ch_hi, w_lo, w_hi, h_lo, h_hi) in size_chart.items():
        if ch_lo <= chest_mm < ch_hi and w_lo <= waist_mm < w_hi and h_lo <= hip_mm < h_hi:
            candidates.append(label)
    if len(candidates) == 1:
        return candidates[0], False
    elif len(candidates) > 1:
        return candidates[0], True   # between-sizes — flag for UI
    return None, True                # no match — flag for manual selection

dims = response["body_dimensions"]
size, between = recommend_size(
    dims["chest_circumference"]["value"],
    dims["waist_circumference_natural"]["value"],
    dims["hip_circumference"]["value"],
    SIZE_CHART_FEMALE,
)

Handling Size Boundary Cases with range_95

When a customer’s predicted dimension lands near a size boundary, use the 95% prediction interval to surface the uncertainty:

def size_with_range_check(dim_key, response):
    dim = response["body_dimensions"][dim_key]
    value    = dim["value"]
    lower_95 = dim["range_95"][0]
    upper_95 = dim["range_95"][1]
    # If range_95 spans two size thresholds → between_sizes = True
    # UI: "We recommend size M — if you prefer a relaxed fit, also consider L."

Response Handling Tips

• Map chest_circumference, waist_circumference_natural, and hip_circumference against your size chart thresholds. For dresses and one-piece garments, the governing dimension is typically hip_circumference.
• Use range_95 for boundary handling: if the 95% interval spans two size thresholds, present both options and let the customer choose based on preferred fit (snug vs. relaxed).
• Check biological_limit_status before mapping: dimensions flagged "OUT_OF_BOUNDS" fall outside population normal range and should not drive automated size recommendations.
• type: "MEASURED" dimensions carry confidence_score: 100 and are returned as-is — the engine never overwrites them.
• For trouser recommendations in a single API call, use bundle: "FULL_BODY" or add "inseam_length", "thigh_circumference" to specific_dimensions.
• When exact_age is not collected, omit the field entirely — the engine defaults to ADULT. Providing exact_age improves the body-composition correction used internally.