import random
import os
import numpy as np
import glob
import matplotlib.pyplot as plt
from collections import Counter
from PIL import Image, ImageDraw, ImageFont
from tqdm import tqdm
from ultralytics.utils.plotting import Annotator, colors
images_path = "./images"
labels_path = "./labels"
images = glob.glob(os.path.join(images_path, "*.jpg")) + \
glob.glob(os.path.join(images_path, "*.JPG"))
label_map = {
0: "Illustration",
1: "Initial",
2: "Ornament",
3: "Stamp",
4: "Table",
}
output_dir = "./generated_html"
if os.path.isdir(output_dir):
print(f"{output_dir} existe déjà")
else:
os.mkdir(output_dir)
annotations_dir = "./annotations"
if os.path.isdir(annotations_dir):
print(f"{annotations_dir} existe déjà")
else:
os.mkdir(annotations_dir)
def classes_visualisation(labels_path: str, label_map: dict, output_dir: str):
total_files = 0
total_labels = []
for filename in os.listdir(labels_path):
if not filename.endswith(".txt"):
continue
total_files += 1
input_path = os.path.join(labels_path, filename)
with open(input_path, "r") as f:
lines = f.readlines()
for line in lines:
parts = line.strip().split()
if not parts:
continue
label = int(parts[0])
total_labels.append(label)
counts = Counter(total_labels)
labels = [label_map[k] for k in counts.keys()]
values = list(counts.values())
total_count = sum(values)
# Fonction pour afficher pourcentage + valeur absolue dans chaque part
def make_autopct(values):
def my_autopct(pct):
absolute = int(round(pct / 100.0 * sum(values)))
return f"{pct:.1f}%\n({absolute})"
return my_autopct
plt.figure(figsize=(7, 7))
plt.pie(
values,
labels=labels,
autopct=make_autopct(values),
textprops={"fontsize": 9},
)
plt.title(
f"GenHisDoc classes distribution\n"
f"Total files: {total_files} | Total labels: {total_count}"
)
# Légende avec le détail des effectifs par classe
legend_labels = [f"{lab} (n={val})" for lab, val in zip(labels, values)]
plt.legend(
legend_labels,
title="Classes",
loc="center left",
bbox_to_anchor=(1, 0, 0.5, 1),
)
plt.tight_layout()
plt.savefig(
os.path.join(output_dir, "GenHisDoc_class_distribution.png"),
bbox_inches="tight",
)
plt.close()
with open(f"{output_dir}/index.html", "w") as f:
f.write(
"""
"""
)
def draw_yolo_annotations(image_path: str, label_path: str, label_map: dict) -> Image.Image | None:
"""Dessine les bounding boxes YOLO sur l'image et retourne une PIL Image."""
if not os.path.exists(image_path):
print(f"Image introuvable : {image_path}")
return None
if not os.path.exists(label_path):
print(f"Label introuvable : {label_path}")
return None
img = np.array(Image.open(image_path).convert("RGB"))
h, w = img.shape[:2]
annotator = Annotator(img, line_width=2)
with open(label_path, "r") as f:
for line in f:
parts = line.strip().split()
if len(parts) < 5:
continue
cls_id = int(parts[0])
cx, cy, bw, bh = map(float, parts[1:5])
# Conversion YOLO (normalisé) → pixels (x1, y1, x2, y2)
x1 = int((cx - bw / 2) * w)
y1 = int((cy - bh / 2) * h)
x2 = int((cx + bw / 2) * w)
y2 = int((cy + bh / 2) * h)
label = label_map.get(cls_id, str(cls_id))
annotator.box_label([x1, y1, x2, y2], label=label, color=colors(cls_id, True))
result = annotator.result()
return Image.fromarray(result)
def controle(label_map: dict):
images_dir = images_path
labels_dir = labels_path
identifier_list = []
annotations_crées = 0
annotations_ignorées = 0
print("génération des annotations")
for filename in tqdm(os.listdir(labels_dir)):
if not filename.endswith(".txt"):
continue
identifier = filename.replace(".txt", "")
identifier_list.append(identifier)
output_path = os.path.join(annotations_dir, f"{identifier}.jpg")
image = draw_yolo_annotations(
os.path.join(images_dir, f"{identifier}.jpg"),
os.path.join(labels_dir, f"{identifier}.txt"),
label_map,
)
if image is None:
annotations_ignorées += 1
continue
if not os.path.isfile(output_path):
image.save(output_path)
annotations_crées += 1
else:
annotations_ignorées += 1
print(f"Annotations créées : {annotations_crées}")
print(f"Annotations ignorées : {annotations_ignorées}")
classes_visualisation(labels_path, label_map, output_dir)
controle(label_map)