Server error 500 using ai-vision detectLivenessWithVerify API

Question

When we try to use detectLivenessWithVerify with the given payload:

const body = [
        {
          name: "verifyImage" as const,
          body: referenceImage, // Buffer
        },
        {
          name: "livenessOperationMode" as const,
          body: "PassiveActive",
        },
        {
          name: "deviceCorrelationId" as const,
          body: deviceId,
        },
        {
          name: "enableSessionImage" as const,
          body: true,
        },
      ];

We get this error:
HTTP 500

{
    code: "Unspecified"
    message: "Internal server error"
}

Answer 1

Hello Balázs Baumgartner,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you are having Server error 500 using ai-vision detectLivenessWithVerify API.

To resolve the HTTP 500 error when using detectLivenessWithVerify, you will need to follow these steps:

A. Ensure you're using multipart/form-data correctly to addresses the original problem. If using JavaScript/TypeScript with fetch or axios, structure it like this:

const formData = new FormData();
formData.append(
  "verifyImage",
  new Blob([referenceImage], { type: "image/jpeg" }), // Or image/png
  "reference.jpg"
);

formData.append("livenessOperationMode", "Passive"); // "Active" also supported
formData.append("deviceCorrelationId", deviceId);
formData.append("enableSessionImage", "true");
const response = await fetch("https://<your-region>.api.cognitive.microsoft.com/vision/v3.2/detectLivenessWithVerify", {
  method: "POST",
  headers: {
    "Ocp-Apim-Subscription-Key": "<your-subscription-key>"
    // Do NOT set Content-Type here; browser will set correct multipart boundaries
  },
  body: formData,
});
const result = await response.json();
console.log(result);

B. According to Microsoft Learn on this tutorial - https://learn.microsoft.com/en-us/azure/ai-services/computer-vision/tutorials/liveness , valid values are:

• "Passive"
• "Active"

Avoid using "PassiveActive" unless explicitly supported in your API version.

C. Confirm Image Format and Size

• Must be JPEG or PNG
• Recommended size:< 4MB
• Must be a valid image buffer

D. Ensure you're calling the correct endpoint, such as POST https://<your-region>.api.cognitive.microsoft.com/vision/v3.2/detectLivenessWithVerify

And using the correct API version (e.g., v3.2 or latest supported).

E. If using SDKs, enable logging to capture the full request/response. This often reveals hidden issues like:

• Invalid headers
• Malformed multipart boundaries
• Authentication failures

Other debugging tips after the above are the followings:

• Test with a known valid image, under 4MB in JPEG/PNG.
• Avoid manually setting Content-Type when using FormData. The browser/Fetch API handles it.
• Make sure referenceImage is a valid Uint8Array or Buffer. If using Node.js, wrap it properly in Blob.
• Confirm endpoint: https://<region>.api.cognitive.microsoft.com/vision/v3.2/detectLivenessWithVerify

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.