Unable to reproduce Qwen2-VL-2B-Instruct result on Hailo10H

Dear Hailo Community, I am trying to reproduce this example from Hugging Face using Qwen2-VL-2B-Instruct for zero-shot object detection. I modified the simple_vlm_chat example from Hailo-apps and fed the API the same prompts and images as the HuggingFace example . prompt = [ { “role”: “user”, “content”: [ {“type”: “image”}, {“type”: “text”, “text”: ‘You are a helpfull assistant to detect objects in images.