Datasets:
ArXiv:
License:
Jules Musquin commited on
Commit ·
c6e077b
1
Parent(s): 2839cb6
update on html_generator
Browse files- .gitignore +3 -1
- .python-version +1 -0
- html_generator.py +180 -0
- main.py +6 -0
- pyproject.toml +11 -0
- requirement.txt +2 -2
- split.py +0 -1
- uv.lock +0 -0
- visualisation.py +0 -80
.gitignore
CHANGED
|
@@ -8,6 +8,7 @@ share/
|
|
| 8 |
log/
|
| 9 |
lib64
|
| 10 |
pyvenv.cfg
|
|
|
|
| 11 |
# Variables d'environnement — NE JAMAIS commiter ce fichier !
|
| 12 |
.env
|
| 13 |
|
|
@@ -23,4 +24,5 @@ lib/
|
|
| 23 |
lib64/
|
| 24 |
|
| 25 |
#website
|
| 26 |
-
generated_html/
|
|
|
|
|
|
| 8 |
log/
|
| 9 |
lib64
|
| 10 |
pyvenv.cfg
|
| 11 |
+
|
| 12 |
# Variables d'environnement — NE JAMAIS commiter ce fichier !
|
| 13 |
.env
|
| 14 |
|
|
|
|
| 24 |
lib64/
|
| 25 |
|
| 26 |
#website
|
| 27 |
+
generated_html/
|
| 28 |
+
annotations/
|
.python-version
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
3.13
|
html_generator.py
ADDED
|
@@ -0,0 +1,180 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import random
|
| 2 |
+
import os
|
| 3 |
+
import numpy as np
|
| 4 |
+
import glob
|
| 5 |
+
import matplotlib.pyplot as plt
|
| 6 |
+
|
| 7 |
+
from collections import Counter
|
| 8 |
+
from PIL import Image, ImageDraw, ImageFont
|
| 9 |
+
from tqdm import tqdm
|
| 10 |
+
from ultralytics.utils.plotting import Annotator, colors
|
| 11 |
+
|
| 12 |
+
images_path = "./images"
|
| 13 |
+
labels_path = "./labels"
|
| 14 |
+
|
| 15 |
+
images = glob.glob(os.path.join(images_path, "*.jpg")) + \
|
| 16 |
+
glob.glob(os.path.join(images_path, "*.JPG"))
|
| 17 |
+
|
| 18 |
+
label_map = {
|
| 19 |
+
0: "Illustration",
|
| 20 |
+
1: "Initial",
|
| 21 |
+
2: "Ornament",
|
| 22 |
+
3: "Stamp",
|
| 23 |
+
4: "Table",
|
| 24 |
+
}
|
| 25 |
+
|
| 26 |
+
output_dir = "./generated_html"
|
| 27 |
+
if os.path.isdir(output_dir):
|
| 28 |
+
print(f"{output_dir} existe déjà")
|
| 29 |
+
else:
|
| 30 |
+
os.mkdir(output_dir)
|
| 31 |
+
|
| 32 |
+
annotations_dir = "./annotations"
|
| 33 |
+
if os.path.isdir(annotations_dir):
|
| 34 |
+
print(f"{annotations_dir} existe déjà")
|
| 35 |
+
else:
|
| 36 |
+
os.mkdir(annotations_dir)
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
def classes_visualisation(labels_path: str, label_map: dict, output_dir: str):
|
| 40 |
+
total_files = 0
|
| 41 |
+
total_labels = []
|
| 42 |
+
|
| 43 |
+
for filename in os.listdir(labels_path):
|
| 44 |
+
if not filename.endswith(".txt"):
|
| 45 |
+
continue
|
| 46 |
+
|
| 47 |
+
total_files += 1
|
| 48 |
+
input_path = os.path.join(labels_path, filename)
|
| 49 |
+
|
| 50 |
+
with open(input_path, "r") as f:
|
| 51 |
+
lines = f.readlines()
|
| 52 |
+
|
| 53 |
+
for line in lines:
|
| 54 |
+
parts = line.strip().split()
|
| 55 |
+
if not parts:
|
| 56 |
+
continue
|
| 57 |
+
label = int(parts[0])
|
| 58 |
+
total_labels.append(label)
|
| 59 |
+
|
| 60 |
+
counts = Counter(total_labels)
|
| 61 |
+
labels = [label_map[k] for k in counts.keys()]
|
| 62 |
+
values = list(counts.values())
|
| 63 |
+
total_count = sum(values)
|
| 64 |
+
|
| 65 |
+
# Fonction pour afficher pourcentage + valeur absolue dans chaque part
|
| 66 |
+
def make_autopct(values):
|
| 67 |
+
def my_autopct(pct):
|
| 68 |
+
absolute = int(round(pct / 100.0 * sum(values)))
|
| 69 |
+
return f"{pct:.1f}%\n({absolute})"
|
| 70 |
+
return my_autopct
|
| 71 |
+
|
| 72 |
+
plt.figure(figsize=(7, 7))
|
| 73 |
+
plt.pie(
|
| 74 |
+
values,
|
| 75 |
+
labels=labels,
|
| 76 |
+
autopct=make_autopct(values),
|
| 77 |
+
textprops={"fontsize": 9},
|
| 78 |
+
)
|
| 79 |
+
plt.title(
|
| 80 |
+
f"GenHisDoc classes distribution\n"
|
| 81 |
+
f"Total files: {total_files} | Total labels: {total_count}"
|
| 82 |
+
)
|
| 83 |
+
|
| 84 |
+
# Légende avec le détail des effectifs par classe
|
| 85 |
+
legend_labels = [f"{lab} (n={val})" for lab, val in zip(labels, values)]
|
| 86 |
+
plt.legend(
|
| 87 |
+
legend_labels,
|
| 88 |
+
title="Classes",
|
| 89 |
+
loc="center left",
|
| 90 |
+
bbox_to_anchor=(1, 0, 0.5, 1),
|
| 91 |
+
)
|
| 92 |
+
|
| 93 |
+
plt.tight_layout()
|
| 94 |
+
plt.savefig(
|
| 95 |
+
os.path.join(output_dir, "GenHisDoc_class_distribution.png"),
|
| 96 |
+
bbox_inches="tight",
|
| 97 |
+
)
|
| 98 |
+
plt.close()
|
| 99 |
+
|
| 100 |
+
with open(f"{output_dir}/index.html", "w") as f:
|
| 101 |
+
f.write(
|
| 102 |
+
"""<!DOCTYPE html>
|
| 103 |
+
<html>
|
| 104 |
+
<head>
|
| 105 |
+
<meta charset="UTF-8">
|
| 106 |
+
<link type="text/css" rel="stylesheet" href="style.css">
|
| 107 |
+
</head>
|
| 108 |
+
<body>
|
| 109 |
+
<header>
|
| 110 |
+
</header>
|
| 111 |
+
<img src="./GenHisDoc_class_distribution.png">
|
| 112 |
+
</body>
|
| 113 |
+
</html>
|
| 114 |
+
"""
|
| 115 |
+
)
|
| 116 |
+
|
| 117 |
+
def draw_yolo_annotations(image_path: str, label_path: str, label_map: dict) -> Image.Image | None:
|
| 118 |
+
"""Dessine les bounding boxes YOLO sur l'image et retourne une PIL Image."""
|
| 119 |
+
if not os.path.exists(image_path):
|
| 120 |
+
print(f"Image introuvable : {image_path}")
|
| 121 |
+
return None
|
| 122 |
+
|
| 123 |
+
if not os.path.exists(label_path):
|
| 124 |
+
print(f"Label introuvable : {label_path}")
|
| 125 |
+
return None
|
| 126 |
+
img = np.array(Image.open(image_path).convert("RGB"))
|
| 127 |
+
h, w = img.shape[:2]
|
| 128 |
+
annotator = Annotator(img, line_width=2)
|
| 129 |
+
|
| 130 |
+
with open(label_path, "r") as f:
|
| 131 |
+
for line in f:
|
| 132 |
+
parts = line.strip().split()
|
| 133 |
+
if len(parts) < 5:
|
| 134 |
+
continue
|
| 135 |
+
cls_id = int(parts[0])
|
| 136 |
+
cx, cy, bw, bh = map(float, parts[1:5])
|
| 137 |
+
# Conversion YOLO (normalisé) → pixels (x1, y1, x2, y2)
|
| 138 |
+
x1 = int((cx - bw / 2) * w)
|
| 139 |
+
y1 = int((cy - bh / 2) * h)
|
| 140 |
+
x2 = int((cx + bw / 2) * w)
|
| 141 |
+
y2 = int((cy + bh / 2) * h)
|
| 142 |
+
label = label_map.get(cls_id, str(cls_id))
|
| 143 |
+
annotator.box_label([x1, y1, x2, y2], label=label, color=colors(cls_id, True))
|
| 144 |
+
|
| 145 |
+
result = annotator.result()
|
| 146 |
+
return Image.fromarray(result)
|
| 147 |
+
|
| 148 |
+
def controle(label_map: dict):
|
| 149 |
+
images_dir = images_path
|
| 150 |
+
labels_dir = labels_path
|
| 151 |
+
identifier_list = []
|
| 152 |
+
annotations_crées = 0
|
| 153 |
+
annotations_ignorées = 0
|
| 154 |
+
print("génération des annotations")
|
| 155 |
+
|
| 156 |
+
for filename in tqdm(os.listdir(labels_dir)):
|
| 157 |
+
if not filename.endswith(".txt"):
|
| 158 |
+
continue
|
| 159 |
+
identifier = filename.replace(".txt", "")
|
| 160 |
+
identifier_list.append(identifier)
|
| 161 |
+
output_path = os.path.join(annotations_dir, f"{identifier}.jpg")
|
| 162 |
+
image = draw_yolo_annotations(
|
| 163 |
+
os.path.join(images_dir, f"{identifier}.jpg"),
|
| 164 |
+
os.path.join(labels_dir, f"{identifier}.txt"),
|
| 165 |
+
label_map,
|
| 166 |
+
)
|
| 167 |
+
if image is None:
|
| 168 |
+
annotations_ignorées += 1
|
| 169 |
+
continue
|
| 170 |
+
if not os.path.isfile(output_path):
|
| 171 |
+
image.save(output_path)
|
| 172 |
+
annotations_crées += 1
|
| 173 |
+
else:
|
| 174 |
+
annotations_ignorées += 1
|
| 175 |
+
|
| 176 |
+
print(f"Annotations créées : {annotations_crées}")
|
| 177 |
+
print(f"Annotations ignorées : {annotations_ignorées}")
|
| 178 |
+
|
| 179 |
+
classes_visualisation(labels_path, label_map, output_dir)
|
| 180 |
+
#controle(label_map)
|
main.py
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
def main():
|
| 2 |
+
print("Hello from genhisdoc-dataset!")
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
if __name__ == "__main__":
|
| 6 |
+
main()
|
pyproject.toml
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[project]
|
| 2 |
+
name = "genhisdoc-dataset"
|
| 3 |
+
version = "0.1.0"
|
| 4 |
+
description = "Add your description here"
|
| 5 |
+
readme = "README.md"
|
| 6 |
+
requires-python = ">=3.13"
|
| 7 |
+
dependencies = [
|
| 8 |
+
"matplotlib>=3.11.0",
|
| 9 |
+
"tqdm>=4.68.4",
|
| 10 |
+
"ultralytics>=8.4.92",
|
| 11 |
+
]
|
requirement.txt
CHANGED
|
@@ -34,7 +34,6 @@ nvidia-nvshmem-cu13==3.4.5
|
|
| 34 |
nvidia-nvtx==13.0.85
|
| 35 |
opencv-python==5.0.0.93
|
| 36 |
packaging==26.2
|
| 37 |
-
pathlib==1.0.1
|
| 38 |
pillow==12.3.0
|
| 39 |
polars==1.42.1
|
| 40 |
polars-runtime-32==1.42.1
|
|
@@ -48,8 +47,9 @@ six==1.17.0
|
|
| 48 |
sympy==1.14.0
|
| 49 |
torch==2.13.0
|
| 50 |
torchvision==0.28.0
|
|
|
|
| 51 |
triton==3.7.1
|
| 52 |
typing_extensions==4.16.0
|
| 53 |
-
ultralytics==8.4.
|
| 54 |
ultralytics-thop==2.0.20
|
| 55 |
urllib3==2.7.0
|
|
|
|
| 34 |
nvidia-nvtx==13.0.85
|
| 35 |
opencv-python==5.0.0.93
|
| 36 |
packaging==26.2
|
|
|
|
| 37 |
pillow==12.3.0
|
| 38 |
polars==1.42.1
|
| 39 |
polars-runtime-32==1.42.1
|
|
|
|
| 47 |
sympy==1.14.0
|
| 48 |
torch==2.13.0
|
| 49 |
torchvision==0.28.0
|
| 50 |
+
tqdm==4.68.4
|
| 51 |
triton==3.7.1
|
| 52 |
typing_extensions==4.16.0
|
| 53 |
+
ultralytics==8.4.92
|
| 54 |
ultralytics-thop==2.0.20
|
| 55 |
urllib3==2.7.0
|
split.py
CHANGED
|
@@ -9,7 +9,6 @@ import random
|
|
| 9 |
current_dir = "./images"
|
| 10 |
val_pct = 10 # 10% validation
|
| 11 |
test_pct = 10 # 10% test
|
| 12 |
-
# le reste (80%) va dans train
|
| 13 |
|
| 14 |
# Récupère toutes les images (en gérant .jpg et .JPG)
|
| 15 |
images = glob.glob(os.path.join(current_dir, "*.jpg")) + \
|
|
|
|
| 9 |
current_dir = "./images"
|
| 10 |
val_pct = 10 # 10% validation
|
| 11 |
test_pct = 10 # 10% test
|
|
|
|
| 12 |
|
| 13 |
# Récupère toutes les images (en gérant .jpg et .JPG)
|
| 14 |
images = glob.glob(os.path.join(current_dir, "*.jpg")) + \
|
uv.lock
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
visualisation.py
DELETED
|
@@ -1,80 +0,0 @@
|
|
| 1 |
-
from pathlib import Path
|
| 2 |
-
from PIL import Image, ImageDraw, ImageFont
|
| 3 |
-
from tqdm import tqdm
|
| 4 |
-
import random
|
| 5 |
-
import os
|
| 6 |
-
import numpy as np
|
| 7 |
-
from ultralytics.utils.plotting import Annotator, colors
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
def draw_yolo_annotations(image_path: Path, label_path: Path, label_map: dict) -> Image.Image | None:
|
| 11 |
-
"""Dessine les bounding boxes YOLO sur l'image et retourne une PIL Image."""
|
| 12 |
-
if not image_path.exists():
|
| 13 |
-
print(f"Image introuvable : {image_path}")
|
| 14 |
-
return None
|
| 15 |
-
if not label_path.exists():
|
| 16 |
-
print(f"Label introuvable : {label_path}")
|
| 17 |
-
return None
|
| 18 |
-
img = np.array(Image.open(image_path).convert("RGB"))
|
| 19 |
-
h, w = img.shape[:2]
|
| 20 |
-
annotator = Annotator(img, line_width=2)
|
| 21 |
-
with open(label_path, "r") as f:
|
| 22 |
-
for line in f:
|
| 23 |
-
parts = line.strip().split()
|
| 24 |
-
if len(parts) < 5:
|
| 25 |
-
continue
|
| 26 |
-
cls_id = int(parts[0])
|
| 27 |
-
cx, cy, bw, bh = map(float, parts[1:5])
|
| 28 |
-
# Conversion YOLO (normalisé) → pixels (x1, y1, x2, y2)
|
| 29 |
-
x1 = int((cx - bw / 2) * w)
|
| 30 |
-
y1 = int((cy - bh / 2) * h)
|
| 31 |
-
x2 = int((cx + bw / 2) * w)
|
| 32 |
-
y2 = int((cy + bh / 2) * h)
|
| 33 |
-
label = label_map.get(cls_id, str(cls_id))
|
| 34 |
-
annotator.box_label([x1, y1, x2, y2], label=label, color=colors(cls_id, True))
|
| 35 |
-
|
| 36 |
-
result = annotator.result()
|
| 37 |
-
return Image.fromarray(result)
|
| 38 |
-
|
| 39 |
-
def controle():
|
| 40 |
-
|
| 41 |
-
label_map = {
|
| 42 |
-
0: "Illustration",
|
| 43 |
-
1: "Ornament",
|
| 44 |
-
2: "Initial",
|
| 45 |
-
3: "Stamp",
|
| 46 |
-
4: "Table",
|
| 47 |
-
}
|
| 48 |
-
|
| 49 |
-
images_dir = Path("images")
|
| 50 |
-
labels_dir = Path("labels")
|
| 51 |
-
|
| 52 |
-
identifier_list = []
|
| 53 |
-
annotations_crées = 0
|
| 54 |
-
annotations_ignorées = 0
|
| 55 |
-
|
| 56 |
-
print("génération des annotations")
|
| 57 |
-
|
| 58 |
-
for filename in tqdm(os.listdir(labels_dir)):
|
| 59 |
-
if not filename.endswith(".txt"):
|
| 60 |
-
continue
|
| 61 |
-
identifier = filename.replace(".txt", "")
|
| 62 |
-
output_path = Path(f'annotations/{identifier}.jpg')
|
| 63 |
-
|
| 64 |
-
image = draw_yolo_annotations(
|
| 65 |
-
images_dir / f'{identifier}.jpg',
|
| 66 |
-
labels_dir / f'{identifier}.txt',
|
| 67 |
-
label_map,
|
| 68 |
-
)
|
| 69 |
-
|
| 70 |
-
if image is None:
|
| 71 |
-
annotations_ignorées += 1
|
| 72 |
-
continue # on passe à l'image suivante
|
| 73 |
-
|
| 74 |
-
if not output_path.is_file():
|
| 75 |
-
image.save(output_path)
|
| 76 |
-
annotations_crées += 1
|
| 77 |
-
else:
|
| 78 |
-
annotations_ignorées += 1
|
| 79 |
-
|
| 80 |
-
controle()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|