Spaces:
Sleeping
Sleeping
SonyaX20
commited on
Commit
·
bf191d3
1
Parent(s):
4d0f186
new
Browse files- .gitignore +22 -0
- README.md +29 -2
- app.py +133 -68
.gitignore
ADDED
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Virtual Environment
|
2 |
+
venv/
|
3 |
+
env/
|
4 |
+
.env
|
5 |
+
|
6 |
+
# Python
|
7 |
+
__pycache__/
|
8 |
+
*.py[cod]
|
9 |
+
*$py.class
|
10 |
+
|
11 |
+
# IDE
|
12 |
+
.vscode/
|
13 |
+
.idea/
|
14 |
+
|
15 |
+
# Temporary files
|
16 |
+
temp_image.png
|
17 |
+
*.log
|
18 |
+
|
19 |
+
# Distribution / packaging
|
20 |
+
dist/
|
21 |
+
build/
|
22 |
+
*.egg-info/
|
README.md
CHANGED
@@ -8,7 +8,34 @@ sdk_version: 5.0.1
|
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
license: mit
|
11 |
-
short_description:
|
12 |
---
|
13 |
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
license: mit
|
11 |
+
short_description: 课程幻灯片智能理解助手
|
12 |
---
|
13 |
|
14 |
+
# 课程幻灯片理解助手
|
15 |
+
|
16 |
+
这是一个基于 Gradio + EasyOCR + GPT-4 的课程幻灯片理解工具。
|
17 |
+
|
18 |
+
## 功能特点
|
19 |
+
|
20 |
+
- 支持上传幻灯片图片
|
21 |
+
- 自动识别幻灯片中的文字内容(支持中英文)
|
22 |
+
- 智能解析幻灯片内容并提供讲解
|
23 |
+
- 支持与AI助手进行对话,深入理解课程内容
|
24 |
+
|
25 |
+
## 技术栈
|
26 |
+
|
27 |
+
- Gradio 5.0.1:构建Web界面
|
28 |
+
- EasyOCR:图片文字识别
|
29 |
+
- OpenAI GPT-4:内容理解与对话
|
30 |
+
- Python 3.8+
|
31 |
+
|
32 |
+
## 使用说明
|
33 |
+
|
34 |
+
1. 上传幻灯片图片
|
35 |
+
2. 点击"分析幻灯片"按钮
|
36 |
+
3. 查看识别的文字内容和AI分析结果
|
37 |
+
4. 使用对话框与AI助手交互,提出问题
|
38 |
+
|
39 |
+
## License
|
40 |
+
|
41 |
+
MIT License
|
app.py
CHANGED
@@ -15,107 +15,172 @@ client = OpenAI(api_key=os.getenv('OPENAI_API_KEY'))
|
|
15 |
# 初始化 EasyOCR
|
16 |
reader = easyocr.Reader(['ch_sim', 'en'])
|
17 |
|
18 |
-
def
|
19 |
-
"""
|
20 |
-
if
|
21 |
-
|
22 |
-
else:
|
23 |
-
# 将 numpy array 转换为 PIL Image
|
24 |
-
if isinstance(image, np.ndarray):
|
25 |
-
image = Image.fromarray(image)
|
26 |
-
# 保存临时文件
|
27 |
-
image_path = "temp_image.png"
|
28 |
-
image.save(image_path)
|
29 |
|
30 |
-
#
|
31 |
-
|
|
|
|
|
32 |
|
33 |
-
#
|
34 |
-
|
35 |
-
os.remove(image_path)
|
36 |
|
37 |
-
|
38 |
-
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
|
41 |
def analyze_slide(text):
|
42 |
"""使用 GPT-4 分析幻灯片内容"""
|
43 |
-
|
|
|
44 |
|
45 |
-
{text}
|
46 |
|
47 |
-
|
48 |
-
1.
|
49 |
-
2.
|
50 |
-
3.
|
51 |
-
4.
|
52 |
|
53 |
-
|
54 |
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
|
|
|
|
|
|
|
|
61 |
|
62 |
-
def chat_with_assistant(message, history):
|
63 |
"""与 AI 助手对话"""
|
64 |
-
|
65 |
-
|
66 |
-
]
|
67 |
-
|
68 |
-
# 添加历史对话
|
69 |
-
for human, assistant in history:
|
70 |
-
messages.append({"role": "user", "content": human})
|
71 |
-
messages.append({"role": "assistant", "content": assistant})
|
72 |
|
73 |
-
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
messages=
|
79 |
-
|
80 |
-
|
81 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
82 |
|
83 |
# 创建 Gradio 界面
|
84 |
with gr.Blocks(title="课程幻灯片理解助手") as demo:
|
85 |
-
gr.Markdown("# 课程幻灯片理解助手")
|
86 |
-
gr.Markdown("上传幻灯片图片,AI
|
|
|
|
|
|
|
87 |
|
88 |
with gr.Row():
|
89 |
-
with gr.Column():
|
90 |
-
image_input = gr.Image(
|
91 |
-
|
|
|
|
|
|
|
92 |
|
93 |
-
with gr.Column():
|
94 |
-
text_output = gr.Textbox(
|
95 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
96 |
|
97 |
gr.Markdown("---")
|
98 |
-
gr.Markdown("### 与 AI 助手对话")
|
99 |
-
chatbot = gr.Chatbot(
|
100 |
-
|
101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
102 |
|
103 |
# 设置事件处理
|
104 |
-
|
105 |
-
fn=
|
106 |
inputs=[image_input],
|
107 |
outputs=[text_output, analysis_output]
|
|
|
|
|
|
|
|
|
108 |
)
|
109 |
|
110 |
msg.submit(
|
111 |
fn=chat_with_assistant,
|
112 |
-
inputs=[msg, chatbot],
|
113 |
outputs=[chatbot],
|
114 |
clear_input=True
|
115 |
)
|
116 |
|
117 |
-
clear.click(lambda:
|
118 |
|
119 |
# 启动应用
|
120 |
if __name__ == "__main__":
|
121 |
-
demo.launch()
|
|
|
15 |
# 初始化 EasyOCR
|
16 |
reader = easyocr.Reader(['ch_sim', 'en'])
|
17 |
|
18 |
+
def process_image(image):
|
19 |
+
"""处理上传的图片并返回识别结果和分析"""
|
20 |
+
if image is None:
|
21 |
+
return "请上传图片", "等待图片上传..."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
+
# 提取文字
|
24 |
+
text = extract_text_from_image(image)
|
25 |
+
if not text.strip():
|
26 |
+
return "未能识别到文字内容,请尝试上传清晰的图片", "无法分析空白内容"
|
27 |
|
28 |
+
# 分析内容
|
29 |
+
analysis = analyze_slide(text)
|
|
|
30 |
|
31 |
+
return text, analysis
|
32 |
+
|
33 |
+
def extract_text_from_image(image):
|
34 |
+
"""从图片中提取文字"""
|
35 |
+
try:
|
36 |
+
if isinstance(image, str):
|
37 |
+
image_path = image
|
38 |
+
else:
|
39 |
+
if isinstance(image, np.ndarray):
|
40 |
+
image = Image.fromarray(image)
|
41 |
+
image_path = "temp_image.png"
|
42 |
+
image.save(image_path)
|
43 |
+
|
44 |
+
# 使用 EasyOCR 识别文字
|
45 |
+
result = reader.readtext(image_path)
|
46 |
+
|
47 |
+
# 删除临时文件
|
48 |
+
if image_path == "temp_image.png" and os.path.exists(image_path):
|
49 |
+
os.remove(image_path)
|
50 |
+
|
51 |
+
# 按照位置排序并组织文字
|
52 |
+
sorted_text = []
|
53 |
+
for (bbox, text, prob) in result:
|
54 |
+
if prob > 0.5: # 只保留置信度大于 0.5 的结果
|
55 |
+
sorted_text.append(text)
|
56 |
+
|
57 |
+
return ' '.join(sorted_text)
|
58 |
+
except Exception as e:
|
59 |
+
return f"图片处理出错: {str(e)}"
|
60 |
|
61 |
def analyze_slide(text):
|
62 |
"""使用 GPT-4 分析幻灯片内容"""
|
63 |
+
try:
|
64 |
+
prompt = f"""请分析以下幻灯片内容,并提供清晰的讲解:
|
65 |
|
66 |
+
内容:{text}
|
67 |
|
68 |
+
请按照以下结构组织回答:
|
69 |
+
1. 主要内容:用2-3句话概括核心内容
|
70 |
+
2. 重点解释:详细解释重要概念和关键点
|
71 |
+
3. 知识延伸:与其他知识的联系
|
72 |
+
4. 应用场景:在实际中的应用示例
|
73 |
|
74 |
+
请用中文回答,语言要通俗易懂。"""
|
75 |
|
76 |
+
response = client.chat.completions.create(
|
77 |
+
model="gpt-4",
|
78 |
+
messages=[{"role": "user", "content": prompt}],
|
79 |
+
temperature=0.7,
|
80 |
+
max_tokens=1000
|
81 |
+
)
|
82 |
+
|
83 |
+
return response.choices[0].message.content
|
84 |
+
except Exception as e:
|
85 |
+
return f"内容分析出错: {str(e)}"
|
86 |
|
87 |
+
def chat_with_assistant(message, history, slide_text):
|
88 |
"""与 AI 助手对话"""
|
89 |
+
if not message:
|
90 |
+
return history
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
|
92 |
+
try:
|
93 |
+
context = f"""当前幻灯片内容:{slide_text}
|
94 |
+
|
95 |
+
请基于以上幻灯片内容,回答用户的问题。如果问题与幻灯片内容无关,也可以回答其他问题。"""
|
96 |
+
|
97 |
+
messages = [
|
98 |
+
{"role": "system", "content": "你是一位专业的课程助教,负责帮助学生理解课程内容。请用清晰易懂的中文回答问题。"},
|
99 |
+
{"role": "user", "content": context}
|
100 |
+
]
|
101 |
+
|
102 |
+
# 添加历史对话
|
103 |
+
for human, assistant in history:
|
104 |
+
messages.append({"role": "user", "content": human})
|
105 |
+
messages.append({"role": "assistant", "content": assistant})
|
106 |
+
|
107 |
+
messages.append({"role": "user", "content": message})
|
108 |
+
|
109 |
+
response = client.chat.completions.create(
|
110 |
+
model="gpt-4",
|
111 |
+
messages=messages,
|
112 |
+
temperature=0.7
|
113 |
+
)
|
114 |
+
|
115 |
+
history.append((message, response.choices[0].message.content))
|
116 |
+
return history
|
117 |
+
except Exception as e:
|
118 |
+
history.append((message, f"回答出错: {str(e)}"))
|
119 |
+
return history
|
120 |
|
121 |
# 创建 Gradio 界面
|
122 |
with gr.Blocks(title="课程幻灯片理解助手") as demo:
|
123 |
+
gr.Markdown("# 📚 课程幻灯片理解助手")
|
124 |
+
gr.Markdown("上传幻灯片图片,AI 将自动识别内容并提供详细讲解")
|
125 |
+
|
126 |
+
# 存储当前识别的文字,用于对话上下文
|
127 |
+
current_text = gr.State("")
|
128 |
|
129 |
with gr.Row():
|
130 |
+
with gr.Column(scale=1):
|
131 |
+
image_input = gr.Image(
|
132 |
+
label="上传幻灯片图片",
|
133 |
+
type="pil",
|
134 |
+
tool="select"
|
135 |
+
)
|
136 |
|
137 |
+
with gr.Column(scale=2):
|
138 |
+
text_output = gr.Textbox(
|
139 |
+
label="识别的文字内容",
|
140 |
+
lines=3,
|
141 |
+
placeholder="上传图片后将显示识别的文字内容..."
|
142 |
+
)
|
143 |
+
analysis_output = gr.Textbox(
|
144 |
+
label="AI 讲解分析",
|
145 |
+
lines=10,
|
146 |
+
placeholder="等待分析结果..."
|
147 |
+
)
|
148 |
|
149 |
gr.Markdown("---")
|
150 |
+
gr.Markdown("### 💬 与 AI 助手对话")
|
151 |
+
chatbot = gr.Chatbot(
|
152 |
+
label="对话历史",
|
153 |
+
height=400,
|
154 |
+
placeholder="在这里可以看到对话历史..."
|
155 |
+
)
|
156 |
+
with gr.Row():
|
157 |
+
msg = gr.Textbox(
|
158 |
+
label="输入你的问题",
|
159 |
+
placeholder="请输入你的问题...",
|
160 |
+
scale=4
|
161 |
+
)
|
162 |
+
clear = gr.Button("🗑️ 清除对话", scale=1)
|
163 |
|
164 |
# 设置事件处理
|
165 |
+
image_input.change(
|
166 |
+
fn=process_image,
|
167 |
inputs=[image_input],
|
168 |
outputs=[text_output, analysis_output]
|
169 |
+
).then(
|
170 |
+
fn=lambda x: x,
|
171 |
+
inputs=[text_output],
|
172 |
+
outputs=[current_text]
|
173 |
)
|
174 |
|
175 |
msg.submit(
|
176 |
fn=chat_with_assistant,
|
177 |
+
inputs=[msg, chatbot, current_text],
|
178 |
outputs=[chatbot],
|
179 |
clear_input=True
|
180 |
)
|
181 |
|
182 |
+
clear.click(lambda: ([], ""), outputs=[chatbot, msg])
|
183 |
|
184 |
# 启动应用
|
185 |
if __name__ == "__main__":
|
186 |
+
demo.launch(share=True)
|