test_1 / app.py
aeresd's picture
Update app.py
896a453 verified
raw
history blame
6.56 kB
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
import torch
import streamlit as st
from PIL import Image
import pytesseract
import pandas as pd
import plotly.express as px
# ✅ Step 1: Emoji 翻译模型(你自己训练的模型)
emoji_model_id = "JenniferHJF/qwen1.5-emoji-finetuned"
emoji_tokenizer = AutoTokenizer.from_pretrained(emoji_model_id, trust_remote_code=True)
emoji_model = AutoModelForCausalLM.from_pretrained(
emoji_model_id,
trust_remote_code=True,
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
).to("cuda" if torch.cuda.is_available() else "cpu")
emoji_model.eval()
# ✅ Step 2: 可选择的冒犯性文本识别模型
model_options = {
"Toxic-BERT": "unitary/toxic-bert",
"Roberta Offensive": "cardiffnlp/twitter-roberta-base-offensive",
"BERT Emotion": "bhadresh-savani/bert-base-go-emotion"
}
# ✅ 页面配置
st.set_page_config(page_title="Emoji Offensive Text Detector", page_icon="🚨", layout="wide")
# ✅ 侧边栏:模型选择
with st.sidebar:
st.header("🧠 Configuration")
selected_model = st.selectbox("Choose classification model", list(model_options.keys()))
selected_model_id = model_options[selected_model]
classifier = pipeline("text-classification", model=selected_model_id, device=0 if torch.cuda.is_available() else -1)
# 初始化历史记录
if "history" not in st.session_state:
st.session_state.history = []
# 分类函数
def classify_emoji_text(text: str):
prompt = f"输入:{text}\n输出:"
input_ids = emoji_tokenizer(prompt, return_tensors="pt").to(emoji_model.device)
with torch.no_grad():
output_ids = emoji_model.generate(**input_ids, max_new_tokens=64, do_sample=False)
decoded = emoji_tokenizer.decode(output_ids[0], skip_special_tokens=True)
translated_text = decoded.split("输出:")[-1].strip() if "输出:" in decoded else decoded.strip()
result = classifier(translated_text)[0]
label = result["label"]
score = result["score"]
reasoning = (
f"The sentence was flagged as '{label}' due to potentially offensive phrases. "
"Consider replacing emotionally charged, ambiguous, or abusive terms."
)
st.session_state.history.append({
"text": text,
"translated": translated_text,
"label": label,
"score": score,
"reason": reasoning
})
return translated_text, label, score, reasoning
# 主页面:输入与分析共存
st.title("🚨 Emoji Offensive Text Detector & Analysis Dashboard")
# 文本输入
st.subheader("1. 输入与分类")
default_text = "你是🐷"
text = st.text_area("Enter sentence with emojis:", value=default_text, height=150)
if st.button("🚦 Analyze Text"):
with st.spinner("🔍 Processing..."):
try:
translated, label, score, reason = classify_emoji_text(text)
st.markdown("**Translated sentence:**")
st.code(translated, language="text")
st.markdown(f"**Prediction:** {label}")
st.markdown(f"**Confidence Score:** {score:.2%}")
st.markdown("**Model Explanation:**")
st.info(reason)
except Exception as e:
st.error(f"❌ An error occurred:\n{e}")
# 图片上传与 OCR
st.markdown("---")
st.subheader("2. 图片 OCR & 分类")
uploaded_file = st.file_uploader("Upload an image (JPG/PNG)", type=["jpg","jpeg","png"])
if uploaded_file:
image = Image.open(uploaded_file)
st.image(image, caption="Uploaded Screenshot", use_column_width=True)
with st.spinner("🧠 Extracting text via OCR..."):
ocr_text = pytesseract.image_to_string(image, lang="chi_sim+eng").strip()
if ocr_text:
st.markdown("**Extracted Text:**")
st.code(ocr_text)
translated, label, score, reason = classify_emoji_text(ocr_text)
st.markdown("**Translated sentence:**")
st.code(translated, language="text")
st.markdown(f"**Prediction:** {label}")
st.markdown(f"**Confidence Score:** {score:.2%}")
st.markdown("**Model Explanation:**")
st.info(reason)
else:
st.info("⚠️ No text detected in the image.")
# 分析仪表盘
st.markdown("---")
st.subheader("3. Violation Analysis Dashboard")
if st.session_state.history:
# 展示历史记录
df = pd.DataFrame(st.session_state.history)
st.markdown("### 🧾 Offensive Terms & Suggestions")
for item in st.session_state.history:
st.markdown(f"- 🔹 **Input:** {item['text']}")
st.markdown(f" - ✨ **Translated:** {item['translated']}")
st.markdown(f" - ❗ **Label:** {item['label']} with **{item['score']:.2%}** confidence")
st.markdown(f" - 🔧 **Suggestion:** {item['reason']}")
# 雷达图
radar_df = pd.DataFrame({
"Category": ["Insult","Abuse","Discrimination","Hate Speech","Vulgarity"],
"Score": [0.7,0.4,0.3,0.5,0.6]
})
radar_fig = px.line_polar(radar_df, r='Score', theta='Category', line_close=True, title="⚠️ Risk Radar by Category")
radar_fig.update_traces(line_color='black')
st.plotly_chart(radar_fig)
# —— 新增:单词级冒犯性相关性分析 —— #
st.markdown("### 🧬 Word-level Offensive Correlation")
# 取最近一次翻译文本,按空格拆分单词
last_translated_text = st.session_state.history[-1]["translated"]
words = last_translated_text.split()
# 对每个单词进行分类并收集分数
word_scores = []
for word in words:
try:
res = classifier(word)[0]
word_scores.append({
"Word": word,
"Label": res["label"],
"Score": res["score"]
})
except Exception:
continue
if word_scores:
word_df = pd.DataFrame(word_scores)
word_df = word_df.sort_values(by="Score", ascending=False).reset_index(drop=True)
max_display = 5
# Streamlit 1.22+ 支持 st.toggle,若版本不支持可改用 checkbox
show_more = st.toggle("Show more words", value=False)
display_df = word_df if show_more else word_df.head(max_display)
# 隐藏边框并渲染 HTML 表格
st.markdown(
display_df.to_html(index=False, border=0),
unsafe_allow_html=True
)
else:
st.info("❕ No word-level analysis available.")
else:
st.info("⚠️ No classification data available yet.")