Describe the task
Currently the stream inference uses a different visualization than our single inference endpoint (the response you get when opening an uploaded image)
I expect this is due to this code,
# Build visualization
vis_frame: np.ndarray = Visualizer.overlay_predictions(frame, prediction_response)
Here prediction_response already contains the same info as the single inference endpoint, including an anomaly_map (Base64-encoded anomaly map image) which we could use for the stream inference as well.
Acceptance Criteria
See task description
Priority
Low
Related Epic
No response
Estimated Time
No response
Current Status
In Progress
Additional Information
No response