Interpret prediction results from image object detection models
Stay organized with collectionsSave and categorize content based on your preferences.
After requesting a prediction, Vertex AI returns results based on your
model's objective. AutoML image object detection prediction responses
return all objects found in an image. Each found object has an annotation (label
and normalized bounding box) with a corresponding confidence score. The bounding
box is written as:
"bboxes": [
[xMin, xMax, yMin, yMax],
...]
WherexMin, xMaxare the minimum and maximum x values andyMin, yMaxare the minimum and maximum y values respectively.
Example batch prediction output
Batch AutoML image object detection prediction responses are stored as
JSON Lines files in Cloud Storage buckets. Each line of the JSON Lines
file
contains all objects found in a single image file. Each found object has
an annotation (label and normalized bounding box) with a corresponding
confidence score.
Note: The following JSON Lines example includes line breaks for
readability. In your JSON Lines files, line breaks are included only after each
each JSON object.
Important:Bounding boxes are specified as:
"bboxes": [
[xMin, xMax, yMin, yMax],
...]
WherexMinandxMaxare the minimum and maximum x values andyMinandyMaxare the minimum and maximum y values respectively.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-09-04 UTC."],[],[],null,["# Interpret prediction results from image object detection models\n\nAfter requesting a prediction, Vertex AI returns results based on your model's objective. AutoML image object detection prediction responses return all objects found in an image. Each found object has an annotation (label and normalized bounding box) with a corresponding confidence score. The bounding box is written as:\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\n`\n\"bboxes\": [\n[xMin, xMax, yMin, yMax],\n...]\n`\nWhere `xMin, xMax` are the minimum and maximum x values and `\nyMin, yMax` are the minimum and maximum y values respectively.\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n#### Example batch prediction output\n\nBatch AutoML image object detection prediction responses are stored as\nJSON Lines files in Cloud Storage buckets. Each line of the JSON Lines\nfile\ncontains all objects found in a single image file. Each found object has\nan annotation (label and normalized bounding box) with a corresponding\nconfidence score.\n| **Note: Zero coordinate values omitted.** When the API detects a coordinate (\"x\" or \"y\") value of 0, ***that coordinate is omitted in the\n| JSON response*** . Thus, a response with a bounding poly around the entire image would be \n| **\\[{},{\"x\": 1,\"y\": 1}\\]** . For more information, see [Method: projects.locations.models.predict](https://cloud.google.com/automl/docs/reference/rest/v1/projects.locations.models/predict#boundingpoly).\n\n\n| **Note**: The following JSON Lines example includes line breaks for\n| readability. In your JSON Lines files, line breaks are included only after each\n| each JSON object.\n\n\u003cbr /\u003e\n\n\n\u003cbr /\u003e\n\n**Important:** Bounding boxes are specified as:\n\n\n`\n\"bboxes\": [\n[xMin, xMax, yMin, yMax],\n...]\n`\nWhere `xMin` and `xMax` are the minimum and maximum x values and `\nyMin` and `yMax` are the minimum and maximum y values respectively.\n\n\u003cbr /\u003e\n\n```\n{\n \"instance\": {\"content\": \"gs://bucket/image.jpg\", \"mimeType\": \"image/jpeg\"},\n \"prediction\": {\n \"ids\": [1, 2],\n \"displayNames\": [\"cat\", \"dog\"],\n \"bboxes\": [\n [0.1, 0.2, 0.3, 0.4],\n [0.2, 0.3, 0.4, 0.5]\n ],\n \"confidences\": [0.7, 0.5]\n }\n}\n```"]]
