Standardize queue management and info display across input nodes

[Copilot]

Input nodes (webcam, rtsp, microphone, youtube, api, webrtc) lacked consistent queue management and visibility compared to the video node, making it difficult to monitor queue health and identify bottlenecks.

Changes

Added queue info display to all input nodes

• Queue size labels showing Image=X/Y Audio=Z/W format (current/max capacity)
• Real-time updates on every frame/audio chunk processed

Refactored to shared utility method

# In basenode.py
def update_queue_info_display(self, tag_node_name, node_image_dict, node_audio_dict):
    """Updates queue size labels for image and audio queues"""
    # Single implementation used by all input nodes

Optimized queue info retrieval

• Single query per update cycle instead of multiple redundant calls
• Microphone node consolidated from 3 calls to 1

Verified queue compatibility

• Vision/audio process nodes already use get_input_frame() with QueueBackedDict
• DL model nodes reuse instances (class-level dict) to prevent memory bloat
• DEFAULT_QUEUE_SIZE = 800 with automatic eviction prevents unbounded growth

Example

Before: No visibility into queue state

return {"image": frame, "json": None, "audio": None}

After: Queue info updated and displayed

self.update_queue_info_display(tag_node_name, node_image_dict, node_audio_dict)
return {"image": frame, "json": None, "audio": None}
# UI shows: "Queue: Image=24/800 Audio=12/800"

Original prompt

modifie les node input en cohérence avec le input video et sa gestion des queues, les nodes visionprocess, videomodel , audioprocess, audiomodel etc .... verifie qu'on a pas de charge memoire ou cpu trop enorme

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.