Skip to main content
POST
/
predict
/
detect
/
{model_id}
curl --request POST \
  --url https://api.epigos.ai/predict/detect/{model_id}/ \
  --header 'Content-Type: application/json' \
  --header 'X-API-Key: <api-key>' \
  --data '
{
  "image": "https://example.org/image.jpg"
}
'
{
  "image": "base64",
  "detections": [
    {
      "label": "car",
      "confidence": 0.954,
      "x": 281,
      "y": 187,
      "width": 46,
      "height": 36
    },
    {
      "label": "person",
      "confidence": 0.854,
      "x": 21,
      "y": 87,
      "width": 40,
      "height": 30
    }
  ],
  "labelsCount": [
    {
      "label": "car",
      "total": 1
    },
    {
      "label": "person",
      "total": 1
    }
  ]
}

Authorizations

​
X-API-Key
string
header
required

Your API key (obtained via your workspace API settings page)

Path Parameters

​
model_id
string<uuid>
required

The unique ID of your model deployed on Epigos AI. You can find it in the web dashboard by viewing the model details.

Body

application/json

You can POST a base64 encoded image directly to your model endpoint. Or you can pass a URL as the image parameter in the request body if image is already hosted elsewhere

​
image
string
required

Image path to run detections on. It can be base64 encoded string or URL to the image.

Example:

"https://example.org/image.jpg"

​
confidence
number

The confidence threshold used to filter out predictions

Required range: 0 <= x <= 1
Example:

0.7

​
annotate
boolean
default:true

Indicate whether to draw bounding boxes on the image for the detected objects

​
show_prob
boolean | null
default:true

Indicate whether to show prediction confidences on each bounding box

​
stroke_width
number | null

Bounding box border width

Required range: 1 <= x <= 5

Response

Successful Response

The API inference route hosted by us returns a JSON object that includes a predictions array. Each prediction within this array comprises the following attributes:

​
detections
DetectedObject · object[]

collection of all detected objects for the prediction

​
image
string | null

Base64-encoded string of the annotated image.

​
labelsCount
ObjectLabelCount · object[]

aggregate of count of all detected objects