读取本地图像数据传递给多模态大模型进行处理并返回结果

读取本地图像数据传递给多模态大模型进行处理并返回结果

importbase64fromopenaiimportOpenAIfromPILimportImageimportioimportos# =========================# 配置# =========================API_BASE="http://192.168.8.221:9024/v1"MODEL_NAME="Qwen2-5-VL-72B-Instruct-quantized-w8a8"IMAGE_PATH="/home/data/project/customer_AAA/ultralytics-yolov5-student/6all/000339.jpg"# ← 替换为你的本地图像路径# =========================# 初始化 OpenAI Client# =========================client=OpenAI(base_url=API_BASE,api_key="EMPTY"# vLLM 不校验 key，填任意值即可)# =========================# 读取并编码图片（Base64）# =========================defload_image_base64(image_path:str)->str:withopen(image_path,"rb")asf:image_bytes=f.read()returnbase64.b64encode(image_bytes).decode("utf-8")image_base64=load_image_base64(IMAGE_PATH)# =========================# 构造 Prompt（严格控制格式）# =========================prompt_text=""" 请根据给定图片进行分析，并严格按照以下格式输出： --- ## 一、场景特点 （分点说明） ## 二、预期检测难点 （分点说明） --- 只输出上述结构，不要输出多余内容。 """# =========================# 调用多模态模型# =========================response=client.chat.completions.create(model=MODEL_NAME,messages=[{"role":"user","content":[{"type":"text","text":prompt_text},{"type":"image_url","image_url":{"url":f"data:image/jpeg;base64,{image_base64}"}}]}],temperature=0.2,max_tokens=800)# =========================# 输出结果# =========================print("="*60)print("模型输出结果：")print(response.choices[0].message.content)