search

valentinfrlch / ha-llmvision

Let Home Assistant see!
Apache License 2.0
187 stars 10 forks source link

llmvision.stream_analyzer. No image input provided #74

Closed Ganfoud closed 1 month ago

Ganfoud commented 1 month ago

Bug Description

When testing the action on a stream in the developper tools, I get the error "llmvision.stream_analyzer. No image input provided"

Version: 1.2.1

Service Call

action: llmvision.stream_analyzer
data:
  include_filename: true
  target_width: 1280
  detail: low
  max_tokens: 100
  temperature: 0.2
  provider: 01JAT1BF7V5DTWSHBQ2T2F7EXT
  image_entity:
    - camera.camera_interieure_rdc
  duration: 10
  message: "Describe what happens in the camera feed "

Logs

Paste relevant logs here

Additional context

valentinfrlch commented 1 month ago

You need to provide the max_frames parameter. I should have marked that as required...

Ganfoud commented 1 month ago

I tried with the max frames parameter but same result

action: llmvision.stream_analyzer
data:
  include_filename: true
  target_width: 1280
  detail: low
  max_tokens: 100
  temperature: 0.2
  provider: 01JAT1BF7V5DTWSHBQ2T2F7EXT
  image_entity:
    - camera.camera_interieure_rdc
  duration: 10
  message: "Describe what happens in the camera feed "
  max_frames: 2
valentinfrlch commented 1 month ago

What provider are you using? I tested your exact config and I cannot replicate the issue. Can you check your logs? Are there any warnings? I think this might be related to https://github.com/valentinfrlch/ha-llmvision/issues/71

Ganfoud commented 1 month ago

I found the issue, it was actually an issue with the openai account, not with the integration, sorry for that and thanks for your help