Spaces:
Running
Running
File size: 3,639 Bytes
727cda5 ba455aa 727cda5 ba455aa 727cda5 ba455aa 727cda5 ba455aa |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 |
from huggingface_hub import hf_hub_download
import streamlit as st
import pandas as pd
import matplotlib.pyplot as plt
import os
import zipfile
import shutil
st.set_page_config(layout="wide")
with st.spinner("Downloading dataset"):
results = hf_hub_download(
repo_id="reducto/rd-tablebench",
filename="rd-tablebench.zip",
repo_type="dataset",
)
def unzip_dataset():
if not os.path.exists("unzipped_dataset"):
os.makedirs("unzipped_dataset")
with st.spinner("Unzipping dataset"):
with zipfile.ZipFile(results, "r") as zip_ref:
zip_ref.extractall("unzipped_dataset")
return "unzipped_dataset/rd-tablebench"
if st.button("Redo Unzip"):
if os.path.exists("unzipped_dataset"):
shutil.rmtree("unzipped_dataset")
st.rerun()
dataset = unzip_dataset()
results = f"{dataset}/providers/scores.csv"
assert os.path.exists(results)
st.html("""
<style>
table {
font-family: arial, sans-serif;
border-collapse: collapse;
white-space: pre;
}
td, th {
border: 1px solid #dddddd;
text-align: left;
padding: 8px;
font-weight: normal;
}
</style>
""")
df = pd.read_csv(results)
if "current_index" not in st.session_state:
st.session_state.current_index = 0
col1, col2, col3 = st.columns([2, 5, 2])
with col1:
st.html("<br/>")
if st.button("⬅️ Previous", use_container_width=True):
if st.session_state.current_index > 0:
st.session_state.current_index -= 1
st.rerun()
# Search box and Go button in col2
with col2:
index_input = st.number_input(
"Index",
label_visibility="hidden",
min_value=0,
max_value=len(df) - 1,
value=st.session_state.current_index,
step=1,
)
if st.button("Go", use_container_width=True):
st.session_state.current_index = int(index_input)
st.rerun()
# Next button in col3
with col3:
st.html("<br/>")
if st.button("Next ➡️", use_container_width=True):
if st.session_state.current_index < len(df) - 1:
st.session_state.current_index += 1
st.rerun()
col1, col2 = st.columns([1, 2])
providers = [
"reducto",
"azure",
"textract",
"gcloud",
"unstructured",
"gpt4o",
"chunkr",
]
with col1:
row = df.iloc[st.session_state.current_index]
# Extract scores
scores = [
row[f"{p}_score"] if row[f"{p}_score"] is not None else 0 for p in providers
]
fig, ax = plt.subplots(figsize=(6, 10))
bars = ax.barh(providers[::-1], scores[::-1])
# Customize plot
ax.set_title("Provider Scores Comparison")
ax.set_ylabel("Providers")
ax.set_xlabel("Scores")
ax.set_xlim(0, 1.1)
for bar in bars:
width = bar.get_width()
ax.text(
width,
bar.get_y() + bar.get_height() / 2.0,
f"{width:.3f}",
ha="left",
va="center",
)
plt.tight_layout()
st.pyplot(fig)
with col2:
image_path = f"{dataset}/_images/{row['pdf_path'].replace('.pdf', '.jpg')}"
st.image(image_path, use_column_width=True)
st.write(row)
st.subheader("Groundtruth")
st.html(f"{dataset}/groundtruth/{row['pdf_path'].replace('.pdf', '.html')}")
st.subheader("Provider Outputs")
for p in providers:
with st.expander(p):
provider_html = (
f"{dataset}/providers/{p}/{row['pdf_path'].replace('.pdf', '.html')}"
)
if os.path.exists(provider_html):
st.html(provider_html)
else:
st.error(f"{p} failed to produce a table output for this image")
|