Bounding Box Area Mismatch

Available

object_detection.bbox_area_mismatch

How closely the model's predicted box sizes match the true object sizes, per class.

Contract

Field	Value
version	1.1.0
metric_type	pointwise
scorer_contract	per_row
direction	lower_is_better
entrypoint	aip_metrics_object_detection.answered.bbox_area_mismatch
target_kind	None

Required columns

• image_id
• label
• predictions

Accepted schemas

[
  {
    "name": "gdi_image_v1",
    "task_types": [
      "detection"
    ]
  }
]

Methodology

1. Deterministic - no LLM. Uses Resaro's cv-eval library (area-based, no IoU matching).
2. Group predicted and ground-truth box areas by class.
3. For each class present in both, compute the relative area error |mean predicted - mean truth| / mean truth.
4. Average the per-class relative errors (0.0 when no class is shared).

Score semantics

Scores capture the relative difference between predicted and true box areas. A low score means box sizes closely match the ground truth; a high score means large size errors. Lower is better.

Worked example

One class with a 10x10 ground-truth box (area 100) and a 12x12 predicted box (area 144) -> relative area error = |144 - 100| / 100 = 0.44.

Configuration schema

{
  "properties": {
    "inverted": {
      "default": false,
      "title": "Inverted",
      "type": "boolean",
      "x-aip-param-role": "scoring_metadata"
    },
    "label_map": {
      "anyOf": [
        {
          "additionalProperties": {
            "type": "integer"
          },
          "type": "object"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Label Map"
    },
    "max": {
      "default": 1.0,
      "title": "Max",
      "type": "number",
      "x-aip-param-role": "scoring_metadata"
    },
    "metric_name": {
      "default": "bbox_area_mismatch",
      "title": "Metric Name",
      "type": "string",
      "x-aip-param-role": "scoring_metadata"
    },
    "min": {
      "default": 0.0,
      "title": "Min",
      "type": "number",
      "x-aip-param-role": "scoring_metadata"
    },
    "weight": {
      "default": 1.0,
      "title": "Weight",
      "type": "number",
      "x-aip-param-role": "scoring_metadata"
    }
  },
  "title": "InitializationConfig",
  "type": "object"
}

Execution

{
  "emits_metric_family": false,
  "function_name": null,
  "max_concurrency": null,
  "processing_kind": "cpu",
  "stream_batch_size": null,
  "timeout_seconds": 60
}

Complete manifest

accepts:
- name: gdi_image_v1
  task_types:
  - detection
config_schema:
  properties:
    inverted:
      default: false
      title: Inverted
      type: boolean
      x-aip-param-role: scoring_metadata
    label_map:
      anyOf:
      - additionalProperties:
          type: integer
        type: object
      - type: 'null'
      default: null
      title: Label Map
    max:
      default: 1.0
      title: Max
      type: number
      x-aip-param-role: scoring_metadata
    metric_name:
      default: bbox_area_mismatch
      title: Metric Name
      type: string
      x-aip-param-role: scoring_metadata
    min:
      default: 0.0
      title: Min
      type: number
      x-aip-param-role: scoring_metadata
    weight:
      default: 1.0
      title: Weight
      type: number
      x-aip-param-role: scoring_metadata
  title: InitializationConfig
  type: object
dependencies: []
description: How closely do the model's box sizes match the true objects?
direction: lower_is_better
display_name: Bounding Box Area Mismatch
entrypoint: aip_metrics_object_detection.answered.bbox_area_mismatch
execution:
  emits_metric_family: false
  function_name: null
  max_concurrency: null
  processing_kind: cpu
  stream_batch_size: null
  timeout_seconds: 60
kind: metric
manifest_version: '1'
max_prompt_slots: 0
metric_metadata:
  class_attribution:
    artifact_type: class_attribution
    version: 1
  methodology:
  - Deterministic - no LLM. Uses Resaro's cv-eval library (area-based, no IoU matching).
  - Group predicted and ground-truth box areas by class.
  - For each class present in both, compute the relative area error |mean predicted
    - mean truth| / mean truth.
  - Average the per-class relative errors (0.0 when no class is shared).
  score_semantics: Scores capture the relative difference between predicted and true
    box areas. A low score means box sizes closely match the ground truth; a high
    score means large size errors. Lower is better.
  summary: How closely the model's predicted box sizes match the true object sizes,
    per class.
  worked_example: One class with a 10x10 ground-truth box (area 100) and a 12x12 predicted
    box (area 144) -> relative area error = |144 - 100| / 100 = 0.44.
metric_type: pointwise
name: object_detection.bbox_area_mismatch
partition_types: []
required_columns:
- image_id
- label
- predictions
required_kinds: []
scorer_contract: per_row
target_kind: null
unsupported_trace_shapes: []
version: 1.1.0