Update app.py

app.py

@@ -0,0 +1,84 @@
+import gradio as gr
+import easyocr
+import numpy as np
+import cv2
+from PIL import Image
+import re
+import json
+
+# Initialisation d'EasyOCR
+reader = easyocr.Reader(['fr', 'en'])
+
+def preprocess_image(pil_image):
+    img = np.array(pil_image)
+    img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
+    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+    gray = cv2.equalizeHist(gray)
+    blurred = cv2.GaussianBlur(gray, (3, 3), 0)
+    coords = np.column_stack(np.where(blurred > 0))
+    angle = cv2.minAreaRect(coords)[-1]
+    angle = -(90 + angle) if angle < -45 else -angle
+    (h, w) = blurred.shape
+    M = cv2.getRotationMatrix2D((w // 2, h // 2), angle, 1.0)
+    deskewed = cv2.warpAffine(blurred, M, (w, h), flags=cv2.INTER_CUBIC, borderMode=cv2.BORDER_REPLICATE)
+    thresh = cv2.adaptiveThreshold(deskewed, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
+    return thresh
+
+def ocr_easyocr(image_np):
+    results = reader.readtext(image_np)
+    texte = "\n".join([text[1] for text in results])
+    return texte
+
+def extract_fields(text):
+    data = {}
+    text = text.upper()
+    patterns = {
+        "nom": r"(NOM)[\s:]+([A-Z\-]+)",
+        "prenom": r"(PRENOM)[\s:]+([A-Z\-]+)",
+        "sexe": r"(SEXE)[\s:]+([FM])",
+        "taille": r"(TAILLE)[\s:]+([0-9,]+\s?M)",
+        "nationalite": r"(NATIONALITE)[\s:]+([A-Z]+)",
+        "date_naissance": r"(\d{2}\s(?:JAN|FEB|MAR|APR|MAI|JUN|JUL|AOU|SEP|OCT|NOV|DEC)\s\d{4})",
+        "numero_id": r"([0-9]{16})",
+        "code_pays": r"\bGIN\b",
+        "nin": r"\b[0-9]{15}\b",
+        "lieu_naissance": r"(NAISSANCE|LIEU)[\s:]+([A-Z\-]+)",
+        "prefecture": r"(PREFECTURE)[\s:]+([A-Z\-]+)",
+        "date_emission": r"(EMISSION)[\s:]+(\d{2}\s\w+\s\d{4})",
+        "date_expiration": r"(EXPIRATION)[\s:]+(\d{2}\s\w+\s\d{4})",
+    }
+
+    for key, pattern in patterns.items():
+        match = re.search(pattern, text)
+        if match:
+            data[key] = match.group(2) if len(match.groups()) > 1 else match.group(1)
+
+    return data
+
+def analyser_carte(recto_img, verso_img):
+    try:
+        recto = preprocess_image(recto_img)
+        verso = preprocess_image(verso_img)
+        text_r = ocr_easyocr(recto)
+        text_v = ocr_easyocr(verso)
+        texte_total = text_r + "\n" + text_v
+        champs = extract_fields(texte_total)
+        return texte_total, champs
+    except Exception as e:
+        return f"Erreur : {str(e)}", {}
+
+interface = gr.Interface(
+    fn=analyser_carte,
+    inputs=[
+        gr.Image(type="pil", label="Recto Carte d'identité"),
+        gr.Image(type="pil", label="Verso Carte d'identité")
+    ],
+    outputs=[
+        gr.Textbox(label="Texte brut extrait"),
+        gr.JSON(label="Champs extraits")
+    ],
+    title="OCRIA - Lecture intelligente de carte d'identité guinéenne",
+    description="Téléversez le recto et le verso d'une carte d'identité pour une extraction automatique des informations par IA.",
+)
+
+interface.launch()