app.py

# -*- coding: utf-8 -*-
"""
鸟类知识科普系统（Qwen3优化版） by [你的名字]
ISOM5240 Group Project
"""
import transformers
import gradio as gr
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer
from PIL import Image
import torch



# 初始化模型（兼容Qwen3）
def init_models():
    # 鸟类分类模型（保持不变）
    classifier = pipeline(
        "image-classification",
        model="chriamue/bird-species-classifier",
        device=0 if torch.cuda.is_available() else -1
    )
    
    # 更新为Qwen3模型（官方支持版本）
    text_generator = pipeline(
        "text-generation",
        model="Qwen/Qwen-7B-Chat",  # 使用官方维护版本
        device_map="auto",
        torch_dtype=torch.bfloat16,
        trust_remote_code=True,     # 必须开启
        model_kwargs={
            "revision": "main",
            "force_download": True  # 替换弃用参数
        }
    )
    
    # 语音合成模型（保持不变）
    tts = pipeline(
        "text-to-speech",
        model="facebook/mms-tts-eng",
        device=0 if torch.cuda.is_available() else -1
    )
    
    return classifier, text_generator, tts

# 生成儿童友好的鸟类描述
def generate_child_friendly_text(bird_name):
    PROMPT = f"""以6-12岁儿童能理解的方式描述{bird_name}：
1. 用比喻手法（如：羽毛像彩虹糖纸）
2. 包含一个趣味冷知识（例如：每天吃相当于自身体重30%的食物）
3. 语句长度不超过15个英文单词
4. 避免使用专业术语"""
    
    response = text_generator(
        PROMPT,
        max_new_tokens=150,
        temperature=0.7,
        do_sample=True
    )
    
    return response[0]['generated_text'].split('\n')[2]

# 主处理流程
def process_image(image):
    try:
        classification = classifier(image)
        bird_name = classification[0]['label']
        description = generate_child_friendly_text(bird_name)
        speech = tts(description, forward_params={"speaker_id": 6})
        
        return {
            "bird_name": bird_name,
            "description": description,
            "audio": speech["audio"]
        }
    except Exception as e:
        return f"处理错误: {str(e)}"

# 初始化模型
classifier, text_generator, tts = init_models()

# 创建Gradio界面
with gr.Blocks(theme=gr.themes.Soft(), css=".gradio-container {max-width: 800px}") as demo:
    gr.Markdown("# 🐦 鸟类知识小课堂（Qwen3版）")
    
    with gr.Row():
        image_input = gr.Image(type="pil", label="上传鸟类图片", height=300)
        audio_output = gr.Audio(label="语音讲解", autoplay=True)
    
    with gr.Column():
        name_output = gr.Textbox(label="识别到的鸟类")
        text_output = gr.Textbox(label="趣味知识", lines=4)
    
    examples = gr.Examples(
        examples=["eagle.jpg", "penguin.jpg", "peacock.jpg"],
        inputs=image_input,
        label="示例图片"
    )
    
    image_input.change(
        process_image,
        inputs=image_input,
        outputs=[name_output, text_output, audio_output]
    )

# 部署配置
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)