🧾 Request Parameters

Parameter	Type	Required	Description
`file`	`file`	✅ Yes (if `base64_str` is not provided)	The input image file to be processed (must be provided as a file, e.g., .jpg, .png).
`base64_str (≤ 1MB)`	`string`	✅ Yes (if `file` is not provided)	Base64-encoded string of the image file (used when sending image data directly in requests).
`text`	`string`	✅ Yes	A list of descriptive keywords (separated by semicolons) specifying the objects to detect in the image (e.g., "cat;dog;car").
`conf_thres`	`float`	✅ Yes	Confidence threshold for object detection. Only detections with confidence greater than or equal to this threshold will be included in the results. Default value is 0.4.
`iou_thres`	`float`	✅ Yes	Intersection over Union (IoU) threshold used for non-maxima suppression (NMS). If the IoU between two bounding boxes is above this threshold, the one with the lower confidence will be discarded. Default value is 0.25.

✅ Successful Response Example

![image](https://aipix.net/static/examples/open-vocabulary-object-detection.webp)

![image](https://aipix.net/static/examples/open-vocabulary-object-detection2.webp)

📤 Response Example:

{
  "code": 200,
  "annotations": [
    {
      "bbox": [
        0,
        302,
        547,
        996
      ],
      "label": "sofa",
      "score": 0.8301814794540405
    },
    {
      "bbox": [
        482,
        23,
        692,
        378
      ],
      "label": "lamp",
      "score": 0.793586254119873
    },
    {
      "bbox": [
        548,
        376,
        817,
        869
      ],
      "label": "table",
      "score": 0.7926247715950012
    }
  ],
  "response_time": 0.406
}

❌ Error Responses

HTTP Code	JSON `code`	Description
422	422	Invalid input parameters.
429	429	Too many requests or timeout.
500	500	Internal server error.

📉 Error Response Format

{
  "code": <error_code>,
  "msg": "<brief error description>"
}

Example:

{
  "code": 429,
  "msg": "Too many requests."
}

📌 Notes

We strictly protect user privacy and do not store any user data.
We strive to minimize input restrictions (e.g., image size, text length), but excessive input volume may result in 429 or 500 errors.
The service operates at the maximum capacity of our servers, and we dynamically scale infrastructure to handle varying request loads.

📬 Support & Custom Deployment

For questions or support, please reach out to our team using the contact page.

We also offer dedicated server deployment for enterprise clients — a private, standalone environment that includes full API functionality with no request limits.

Pricing is based on the number of servers deployed, with each server serving at maximum throughput.

📩 Contact Us

📚 View Our API List