Video Inference API: Status 2 "Object Detection Error" on 8s video

Community Help
1

I’m trying to use the Roboflow Video Inference API to analyze an 8-second video (hosted on a public Cloudinary URL). The video simply shows a woman entering a room. My goal is to inspect the room and check if certain tasks are done.

Issue:
When I submit the video to the API endpoint:

https://api.roboflow.com/videoinfer/?api_key=<YOUR_API_KEY>

the job status returns 2 and says “object detection error.” I’m not sure why this is happening. The video is in a supported format (mp4) and is publicly accessible. I also tried using another video and it also gave the same error.

This is the response I get when checking the status:

[
{
“_id”: “14d34094-2fed-406e-b6c8-b0c06213334d”,
“created_at”: “Mon, 16 Jun 2025 14:21:37 GMT”,
“crop_area”: “full”,
“infer_fps”: 5,
“infer_models”: [
{
“api_key”: “…”,
“infer_errors”: 0,
“infer_success”: 0,
“inference_params”: {
“confidence”: 0.4
},
“inference_type”: “object-detection”,
“model_id”: “unit-walkthrough-validation”,
“model_version”: 4
}
],
“memo”: “xazo2DMYRJOUsCsCOQ9soeJuvIN2”,
“metadata”: “”,
“pubsub_message_id”: “15179134466164301”,
“retries”: 1,
“signed_url”: “https://res.cloudinary.com/dvhgcmpi5/video/upload/v1750083696/lou5qtklzd27cnoiwtp1.mp4”,
“status”: 2,
“status_info”: “object-detection error”,
“updated_at”: “Mon, 16 Jun 2025 14:21:44 GMT”,
“webhook_url”: “”
}
]

-What does status 2 and “object detection error” specifically mean?
-Are there any common reasons why a video that seems valid would fail with this error? Maybe because the duration is too short?
-Is there a way to get more detailed error information from the API or logs?

Request body to start the job:

image
image1300×711 55.9 KB

Any help or suggestions would be appreciated!

  • Project Type: object-detection
  • Operating System & Browser: Windows
  • Project Universe Link or Workspace/Project ID: unit-walkthrough-validation
2

Hi @shabeeb!
I just wanted to let you know that I am looking into this and I will get you an answer. Thank you for contributing to the Roboflow community!

3

Hi @shabeeb!
I did some digging and found some answers for you. We now suggest using Batch Processing for your use case instead of the Video Inference API. Here are the supporting docs to aide in your implementation: Batch Processing

Also, we are currently working on providing more detailed error information from the error logs, API, and documentation.

Have a wonderful day!

1 Like
4

Hi @Ford,

Thank you for your suggestion to use Batch Processing for my use case. I’ve reviewed the documentation and see that Batch Processing is supported via the web UI and CLI, and the docs mention REST API support as well. However, I haven’t been able to find explicit REST API endpoint details or example payloads for creating and managing batch jobs directly via API.

Could you please provide the REST API endpoints or any documentation or references for using Batch Processing programmatically? I have to use the API in n8n cloud.

5

Hi @Shabeeb!
Absolutely! Here is a link to the API Batch Processing documentation which has endpoints in it: RoboFlow Batch Processing API Documentation

6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.