Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
3,425 questions
Azure Face API v1.2 – Liveness Session Stuck in "Running" Even After Successful Video Upload (Tested PUT & POST Upload)
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(96, 79%, 19%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ENK%3C/text%3E%3C/svg%3E)
Niket Kumar Singh
600
Reputation points
Hi Team,
We are evaluating Azure Face API v1.2 Liveness Detection (Passive Mode) for server-side implementation via REST APIs. However, we’re encountering consistent issues where the session remains stuck in the "Running" state, even after successful video upload.
Configuration Details:
- API Version:
v1.2 - Mode: Passive / Passive-Active mode
Region: Tested with both East US and West US
Authentication: Using valid subscription key and authToken (used within 10-minute TTL)
Session Creation: Works fine — POST /detectLiveness-sessions returns sessionId + authToken
Video Upload Attempts:
We've tried both approaches, with consistent 200 OK responses:
PUT /detectLiveness-sessions/{sessionId}/video
POST /session/start?sessionId=...
After successful upload, we poll:
GET /detectLiveness-sessions/{sessionId}
But the response is always:
{
"sessionId": "...",
"authToken": "",
"status": "Running",
"results": {
"attempts": []
}
}
Actions Already Taken:
- Re-encoded the video using recommended
ffmpegparameters:
ffmpeg -i input.mp4 -c:v libx264 -b:v 1M -an -vf scale=640:480 -r 25 output_cleaned.mp4- Confirmed:
-
Content-Type: application/octet-stream -
Authorization: Bearer <authToken>(no subscription key here) - Session ID matches what was received during creation
- Upload happens well within 10-minute authToken TTL
-
- Tested using both East US and West US resources
- Clarification if REST-only flow is unstable compared to SDK (we are not using iOS/Android/Web SDKs — only backend REST calls)
- Any known platform-side issue with certain formats/resolutions, even if re-encoded properly?
- Is there any additional processing step required after video upload that we're missing?
- Confirmed:
Azure AI services
{count} votes
-
Abdelaziz Khajour 235 Reputation points Microsoft Employee2025-04-18T08:48:47.21+00:00 To troubleshoot the issue you are facing with Azure Face API v1.2 Liveness Detection, here are some steps and considerations you can take:
- Check Video Upload Status: - Ensure that the video upload is successful by checking the response status code and any error messages returned. - Verify that the video is uploaded in the correct format and resolution. You can re-encode the video using the recommended ffmpeg parameters as you have already done.
- Polling Session Status: - After a successful video upload, make sure to poll the session status using the
GET /detectLiveness-sessions/{sessionId}endpoint. - If the status remains "Running," it indicates that the processing is still ongoing. You may need to wait for the processing to complete before retrieving the results. - Authentication and Session ID: - Double-check that you are using a valid subscription key and authToken for authentication. - Ensure that the session ID you receive during creation matches the one used for video upload and status polling.
- Backend Logs and Debugging: - Check if there are any backend logs available that can provide more insights into the processing status or any errors encountered during the liveness detection session. - Consider enabling detailed logging in your backend system to track the flow of requests and responses.
- REST API Stability: - Verify if the REST-only flow is stable compared to using SDKs. While SDKs may provide additional functionalities and error handling, the REST API should also work reliably for backend implementations. 6. Platform-Side Issues: - Ensure that the video format and resolution are supported by the Azure Face API. While re-encoding the video is a good practice, make sure the output format meets the API requirements. - Check if there are any known issues with certain video formats or resolutions that could affect the processing.
- Additional Processing Steps: - After the video upload, there may be additional processing steps required based on the API documentation. Make sure you are following all the necessary steps post-upload to retrieve the liveness detection results. If the issue persists after following these steps, you may consider reaching out to Azure support for further assistance or collaboration. They can provide real-time support and guidance to help resolve the issue with Azure Face API liveness detection.
I hope these insights help you troubleshoot the issue effectively. If you need further assistance, feel free to ask.
Abdelaziz
-
Niket Kumar Singh 600 Reputation points
2025-04-18T09:30:35.3566667+00:00 Hi Abdelaziz,
Thank you for the detailed guidance.
We have already validated most of the recommendations mentioned, including:
Video is re-encoded with:
ffmpeg -i input.mp4 -c:v libx264 -b:v 1M -an -vf scale=640:480 -r 25 output_cleaned.mp4Both
PUT /detectLiveness-sessions/{sessionId}/videoandPOST /session/start?sessionId=...return200 OK.We poll
GET /detectLiveness-sessions/{sessionId}continuously but it remains in"status": "Running"even after several minutes. Theattemptsarray stays empty.All authentication headers and session IDs are validated, and token is used within the 10-minute TTL.
We’ve tested with Face API resources provisioned in both East US and West US, including the latest model version
"2024-11-15".Currently using only REST APIs for orchestration (no mobile or web SDKs).
We’ve also logged a support request with Microsoft under Tracking ID: #2504160030005702 for deeper investigation.
Would it be possible to connect with a backend engineer or product team over Teams for a short session to confirm whether:
There are any known backend limitations when using REST APIs without SDKs?
The issue is specific to Passive mode?
There's a way to view deeper telemetry/logs tied to a sessionId?
We’re happy to share session IDs or sanitized payloads directly over private channels if that helps with troubleshooting.
Looking forward to your thoughts.
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(6.4, 61%, 10%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ES%3C/text%3E%3C/svg%3E)
santoshkc 14,500 Reputation points Microsoft External Staff Moderator2025-04-22T17:04:48.61+00:00 I hope your issue has been resolved with this support request #2504160030005702. And if you have any further queries do let us know.
Thank you.
-
Niket Kumar Singh 600 Reputation points
2025-04-22T17:46:04.19+00:00 Hi @santoshkc
Not yet the case is going no where, they had open collab with PG team.
If you can help us it will be really apprecited!
-
santoshkc 14,500 Reputation points Microsoft External Staff Moderator
2025-04-24T18:18:14.4833333+00:00 We are reaching out to the PG team to get more information related to your query and will get back to you as soon as we have an update.
-
Niket Kumar Singh 600 Reputation points
2025-04-25T06:03:02.85+00:00 HI santoshkc
As per ms team, there is no api for the liveness detection, we need to use SDK, could you please confirm the same?
-
santoshkc 14,500 Reputation points Microsoft External Staff Moderator
2025-04-28T08:22:52.6733333+00:00 You must use the Azure Face SDKs for Liveness Detection. There is no full API flow supported for production. Thank you.
Sign in to comment
Sign in to answer