Spaces:
Sleeping
Sleeping
Upload 6 files
Browse files- app.py +11 -23
- game_engine.py +78 -132
- image_processing_gpu.py +12 -40
- requirements.txt +6 -9
- utils.py +12 -46
app.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
# ==========================================
|
| 2 |
-
#
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
-
Application principale
|
| 7 |
"""
|
| 8 |
|
| 9 |
import gradio as gr
|
|
@@ -12,29 +12,17 @@ import os
|
|
| 12 |
import gc
|
| 13 |
import numpy as np
|
| 14 |
from PIL import Image
|
| 15 |
-
import simple_gpu
|
| 16 |
|
| 17 |
warnings.filterwarnings("ignore")
|
| 18 |
|
| 19 |
-
# Import
|
| 20 |
-
|
| 21 |
-
import torch
|
| 22 |
-
if torch.cuda.is_available():
|
| 23 |
-
from image_processing_gpu import init_ocr_model, create_white_canvas, cleanup_memory
|
| 24 |
-
print("📱 Interface: Mode GPU détecté - TrOCR")
|
| 25 |
-
else:
|
| 26 |
-
from image_processing_cpu import init_ocr_model, create_white_canvas, cleanup_memory
|
| 27 |
-
print("📱 Interface: Mode CPU détecté - EasyOCR")
|
| 28 |
-
except ImportError:
|
| 29 |
-
from image_processing_cpu import init_ocr_model, create_white_canvas, cleanup_memory
|
| 30 |
-
print("📱 Interface: Mode CPU détecté - EasyOCR")
|
| 31 |
-
|
| 32 |
from game_engine import MathGame, export_to_clean_dataset
|
| 33 |
|
| 34 |
-
print("🚀 Initialisation Calcul OCR v3.0...")
|
| 35 |
-
print("🔄 Chargement modèle
|
| 36 |
init_ocr_model()
|
| 37 |
-
print("✅ Modèle
|
| 38 |
|
| 39 |
def create_new_game_session():
|
| 40 |
"""Crée une nouvelle instance de jeu pour chaque session"""
|
|
@@ -123,7 +111,7 @@ with gr.Blocks(
|
|
| 123 |
head="<meta name='viewport' content='width=device-width, initial-scale=1.0'>"
|
| 124 |
) as demo:
|
| 125 |
|
| 126 |
-
#
|
| 127 |
game_state = gr.State(value=None)
|
| 128 |
|
| 129 |
gr.Markdown(
|
|
@@ -196,7 +184,7 @@ with gr.Blocks(
|
|
| 196 |
export_button = gr.Button("📤 Exporter la série vers le dataset", variant="primary", size="lg")
|
| 197 |
export_status = gr.Markdown("")
|
| 198 |
|
| 199 |
-
#
|
| 200 |
go_button.click(
|
| 201 |
fn=start_game_wrapper,
|
| 202 |
inputs=[duration_choice, operation_choice, difficulty_choice, game_state],
|
|
@@ -216,8 +204,8 @@ with gr.Blocks(
|
|
| 216 |
)
|
| 217 |
|
| 218 |
if __name__ == "__main__":
|
| 219 |
-
print("🚀 Lancement Calcul OCR v3.0 -
|
| 220 |
-
print("
|
| 221 |
print("🎯 Dataset: calcul_ocr_dataset")
|
| 222 |
print("📊 Opérations: ×, +, -, ÷, Aléatoire")
|
| 223 |
print("⚙️ Durées: 30s, 60s")
|
|
|
|
| 1 |
# ==========================================
|
| 2 |
+
# app.py - Calcul OCR v3.0 - Version GPU simplifiée
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
+
Application principale optimisée pour ZeroGPU HuggingFace Spaces
|
| 7 |
"""
|
| 8 |
|
| 9 |
import gradio as gr
|
|
|
|
| 12 |
import gc
|
| 13 |
import numpy as np
|
| 14 |
from PIL import Image
|
|
|
|
| 15 |
|
| 16 |
warnings.filterwarnings("ignore")
|
| 17 |
|
| 18 |
+
# Import GPU uniquement
|
| 19 |
+
from image_processing_gpu import init_ocr_model, create_white_canvas, cleanup_memory
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
from game_engine import MathGame, export_to_clean_dataset
|
| 21 |
|
| 22 |
+
print("🚀 Initialisation Calcul OCR v3.0 (GPU)...")
|
| 23 |
+
print("🔄 Chargement modèle TrOCR...")
|
| 24 |
init_ocr_model()
|
| 25 |
+
print("✅ Modèle TrOCR prêt")
|
| 26 |
|
| 27 |
def create_new_game_session():
|
| 28 |
"""Crée une nouvelle instance de jeu pour chaque session"""
|
|
|
|
| 111 |
head="<meta name='viewport' content='width=device-width, initial-scale=1.0'>"
|
| 112 |
) as demo:
|
| 113 |
|
| 114 |
+
# State management - Chaque utilisateur a sa propre instance
|
| 115 |
game_state = gr.State(value=None)
|
| 116 |
|
| 117 |
gr.Markdown(
|
|
|
|
| 184 |
export_button = gr.Button("📤 Exporter la série vers le dataset", variant="primary", size="lg")
|
| 185 |
export_status = gr.Markdown("")
|
| 186 |
|
| 187 |
+
# Événements avec State management
|
| 188 |
go_button.click(
|
| 189 |
fn=start_game_wrapper,
|
| 190 |
inputs=[duration_choice, operation_choice, difficulty_choice, game_state],
|
|
|
|
| 204 |
)
|
| 205 |
|
| 206 |
if __name__ == "__main__":
|
| 207 |
+
print("🚀 Lancement Calcul OCR v3.0 - Version GPU simplifiée")
|
| 208 |
+
print("🎯 ZeroGPU HuggingFace Spaces")
|
| 209 |
print("🎯 Dataset: calcul_ocr_dataset")
|
| 210 |
print("📊 Opérations: ×, +, -, ÷, Aléatoire")
|
| 211 |
print("⚙️ Durées: 30s, 60s")
|
game_engine.py
CHANGED
|
@@ -1,10 +1,9 @@
|
|
| 1 |
# ==========================================
|
| 2 |
-
# game_engine.py - Calcul OCR v3.0
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
-
Moteur de jeu mathématique
|
| 7 |
-
VERSION MODIFIÉE: Utilise le format Image natif au lieu de Base64
|
| 8 |
"""
|
| 9 |
|
| 10 |
import random
|
|
@@ -20,81 +19,27 @@ import threading
|
|
| 20 |
import queue
|
| 21 |
from typing import Dict, Tuple, Optional
|
| 22 |
|
| 23 |
-
#
|
| 24 |
-
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
-
|
| 28 |
-
space_id = os.getenv("SPACE_ID")
|
| 29 |
-
space_hardware = os.getenv("SPACE_HARDWARE")
|
| 30 |
-
hf_space = os.getenv("HF_SPACE")
|
| 31 |
-
space_author = os.getenv("SPACE_AUTHOR_NAME")
|
| 32 |
-
zero_gpu = os.getenv("ZERO_GPU")
|
| 33 |
|
| 34 |
-
|
| 35 |
-
print(f" SPACE_ID: {space_id}")
|
| 36 |
-
print(f" SPACE_HARDWARE: {space_hardware}")
|
| 37 |
-
print(f" HF_SPACE: {hf_space}")
|
| 38 |
-
print(f" SPACE_AUTHOR_NAME: {space_author}")
|
| 39 |
-
print(f" ZERO_GPU: {zero_gpu}")
|
| 40 |
-
|
| 41 |
-
# Détecter ZeroGPU (GPU dynamique HF)
|
| 42 |
-
is_zerogpu = space_id and ("hoololi" in str(space_id))
|
| 43 |
-
|
| 44 |
-
print(f"🎯 ZeroGPU détecté: {is_zerogpu}")
|
| 45 |
-
|
| 46 |
-
if is_zerogpu:
|
| 47 |
-
try:
|
| 48 |
-
print("🚀 Force mode ZeroGPU - Import GPU...")
|
| 49 |
-
from simple_gpu import gpu_dummy_function
|
| 50 |
-
print("✅ Simple GPU importé")
|
| 51 |
-
|
| 52 |
-
from image_processing_gpu import (
|
| 53 |
-
recognize_number_fast_with_image as gpu_recognize,
|
| 54 |
-
create_thumbnail_fast,
|
| 55 |
-
create_white_canvas,
|
| 56 |
-
cleanup_memory,
|
| 57 |
-
log_memory_usage,
|
| 58 |
-
get_ocr_model_info
|
| 59 |
-
)
|
| 60 |
-
|
| 61 |
-
recognize_number_fast_with_image = gpu_recognize
|
| 62 |
-
ocr_module = "zerogpu_trocr"
|
| 63 |
-
print("✅ Game Engine: Mode ZeroGPU - TrOCR directement utilisé")
|
| 64 |
-
|
| 65 |
-
except Exception as e:
|
| 66 |
-
print(f"❌ Erreur ZeroGPU: {e}")
|
| 67 |
-
from image_processing_cpu import (
|
| 68 |
-
recognize_number_fast_with_image,
|
| 69 |
-
create_thumbnail_fast,
|
| 70 |
-
create_white_canvas,
|
| 71 |
-
cleanup_memory,
|
| 72 |
-
log_memory_usage,
|
| 73 |
-
get_ocr_model_info
|
| 74 |
-
)
|
| 75 |
-
ocr_module = "cpu"
|
| 76 |
-
print("✅ Game Engine: Mode CPU - EasyOCR (fallback)")
|
| 77 |
-
else:
|
| 78 |
-
from image_processing_cpu import (
|
| 79 |
-
recognize_number_fast_with_image,
|
| 80 |
-
create_thumbnail_fast,
|
| 81 |
-
create_white_canvas,
|
| 82 |
-
cleanup_memory,
|
| 83 |
-
log_memory_usage,
|
| 84 |
-
get_ocr_model_info
|
| 85 |
-
)
|
| 86 |
-
ocr_module = "cpu"
|
| 87 |
-
print("✅ Game Engine: Mode CPU - EasyOCR standard")
|
| 88 |
-
|
| 89 |
-
# Récupérer les infos du modèle sélectionné
|
| 90 |
try:
|
| 91 |
ocr_info = get_ocr_model_info()
|
| 92 |
print(f"🎯 OCR sélectionné: {ocr_info['model_name']} sur {ocr_info['device']}")
|
| 93 |
except Exception as e:
|
| 94 |
print(f"⚠️ Impossible de récupérer les infos OCR: {e}")
|
| 95 |
-
ocr_info = {"model_name": "
|
| 96 |
|
| 97 |
-
# Imports dataset
|
| 98 |
try:
|
| 99 |
from datasets import Dataset, Image as DatasetImage, load_dataset
|
| 100 |
DATASET_AVAILABLE = True
|
|
@@ -103,7 +48,7 @@ except ImportError as e:
|
|
| 103 |
DATASET_AVAILABLE = False
|
| 104 |
print(f"⚠️ Modules dataset non disponibles: {e}")
|
| 105 |
|
| 106 |
-
#
|
| 107 |
DATASET_NAME = "hoololi/calcul_ocr_dataset_with_images_v2"
|
| 108 |
|
| 109 |
# Configuration des difficultés par opération
|
|
@@ -170,7 +115,7 @@ def create_result_row_with_images(i: int, image: dict | np.ndarray | Image.Image
|
|
| 170 |
|
| 171 |
|
| 172 |
class MathGame:
|
| 173 |
-
"""Moteur de jeu mathématique avec traitement parallèle"""
|
| 174 |
|
| 175 |
def __init__(self):
|
| 176 |
self.is_running = False
|
|
@@ -222,22 +167,22 @@ class MathGame:
|
|
| 222 |
self.processing_active = True
|
| 223 |
self.worker_thread = threading.Thread(target=self._process_images_worker, daemon=True)
|
| 224 |
self.worker_thread.start()
|
| 225 |
-
print("🔄 Thread de traitement parallèle démarré")
|
| 226 |
|
| 227 |
def _stop_background_processing(self) -> None:
|
| 228 |
"""Arrête le thread de traitement"""
|
| 229 |
self.processing_active = False
|
| 230 |
if self.worker_thread and self.worker_thread.is_alive():
|
| 231 |
-
print("⏹️ Arrêt du thread de traitement parallèle")
|
| 232 |
|
| 233 |
def _process_images_worker(self) -> None:
|
| 234 |
-
"""Worker thread qui traite les images en arrière-plan"""
|
| 235 |
-
print("🚀 Worker thread démarré")
|
| 236 |
while self.processing_active:
|
| 237 |
try:
|
| 238 |
if not self.processing_queue.empty():
|
| 239 |
question_num, image, expected, operation_data = self.processing_queue.get(timeout=1)
|
| 240 |
-
print(f"🔄 Traitement parallèle image {question_num}...")
|
| 241 |
|
| 242 |
start_time = time.time()
|
| 243 |
result_data = create_result_row_with_images(question_num, image, expected, operation_data)
|
|
@@ -245,7 +190,7 @@ class MathGame:
|
|
| 245 |
|
| 246 |
# Stocker le résultat
|
| 247 |
self.results_cache[question_num] = result_data
|
| 248 |
-
print(f"✅ Image {question_num} traitée en {processing_time:.1f}s (parallèle)")
|
| 249 |
|
| 250 |
else:
|
| 251 |
time.sleep(0.1)
|
|
@@ -253,16 +198,16 @@ class MathGame:
|
|
| 253 |
except queue.Empty:
|
| 254 |
continue
|
| 255 |
except Exception as e:
|
| 256 |
-
print(f"❌ Erreur traitement parallèle: {e}")
|
| 257 |
|
| 258 |
-
print("🛑 Worker thread terminé")
|
| 259 |
|
| 260 |
def _add_image_to_processing_queue(self, question_num: int, image: dict | np.ndarray | Image.Image,
|
| 261 |
expected: int, operation_data: tuple) -> None:
|
| 262 |
-
"""Ajoute une image à la queue de traitement"""
|
| 263 |
if image is not None:
|
| 264 |
self.processing_queue.put((question_num, image, expected, operation_data))
|
| 265 |
-
print(f"📝 Image {question_num} ajoutée à la queue de traitement")
|
| 266 |
|
| 267 |
def generate_multiplication(self, difficulty: str) -> tuple[str, int]:
|
| 268 |
"""Génère une multiplication"""
|
|
@@ -358,7 +303,7 @@ class MathGame:
|
|
| 358 |
self.export_timestamp = None
|
| 359 |
self.export_result = None
|
| 360 |
|
| 361 |
-
# Démarrer le traitement parallèle
|
| 362 |
self._start_background_processing()
|
| 363 |
|
| 364 |
gc.collect()
|
|
@@ -406,7 +351,7 @@ class MathGame:
|
|
| 406 |
return self.end_game(image_data)
|
| 407 |
|
| 408 |
if image_data is not None:
|
| 409 |
-
# Ajouter l'image à la liste ET au traitement parallèle
|
| 410 |
self.user_images.append(image_data)
|
| 411 |
self.expected_answers.append(self.correct_answer)
|
| 412 |
|
|
@@ -415,7 +360,7 @@ class MathGame:
|
|
| 415 |
a, op, b = int(parts[0]), parts[1], int(parts[2])
|
| 416 |
current_operation_data = (a, b, op, self.correct_answer)
|
| 417 |
|
| 418 |
-
# Lancer le traitement en parallèle de l'image qu'on vient de recevoir
|
| 419 |
self._add_image_to_processing_queue(self.question_count, image_data, self.correct_answer, current_operation_data)
|
| 420 |
|
| 421 |
self.question_count += 1
|
|
@@ -456,10 +401,10 @@ class MathGame:
|
|
| 456 |
|
| 457 |
self.is_running = False
|
| 458 |
|
| 459 |
-
# Arrêter le traitement parallèle
|
| 460 |
self._stop_background_processing()
|
| 461 |
|
| 462 |
-
print("🏁 Fin de jeu - Assemblage des résultats...")
|
| 463 |
|
| 464 |
if final_image is not None:
|
| 465 |
self.user_images.append(final_image)
|
|
@@ -470,8 +415,8 @@ class MathGame:
|
|
| 470 |
a, op, b = int(parts[0]), parts[1], int(parts[2])
|
| 471 |
final_operation_data = (a, b, op, self.correct_answer)
|
| 472 |
|
| 473 |
-
# Traiter la dernière image immédiatement
|
| 474 |
-
print(f"🔄 Traitement final de l'image {self.question_count}...")
|
| 475 |
final_result = create_result_row_with_images(self.question_count, final_image, self.correct_answer, final_operation_data)
|
| 476 |
self.results_cache[self.question_count] = final_result
|
| 477 |
|
|
@@ -479,17 +424,17 @@ class MathGame:
|
|
| 479 |
if len(self.operations_history) < len(self.user_images):
|
| 480 |
self.operations_history.append((a, b, op, self.correct_answer))
|
| 481 |
|
| 482 |
-
# Attendre que toutes les images soient traitées
|
| 483 |
max_wait = 10
|
| 484 |
wait_start = time.time()
|
| 485 |
expected_results = len(self.user_images)
|
| 486 |
|
| 487 |
-
print(f"⏳ Attente de {expected_results} résultats...")
|
| 488 |
while len(self.results_cache) < expected_results and (time.time() - wait_start) < max_wait:
|
| 489 |
time.sleep(0.1)
|
| 490 |
|
| 491 |
results_ready = len(self.results_cache)
|
| 492 |
-
print(f"✅ {results_ready}/{expected_results} résultats prêts")
|
| 493 |
|
| 494 |
# Assembler les résultats dans l'ordre
|
| 495 |
correct_answers = 0
|
|
@@ -502,14 +447,14 @@ class MathGame:
|
|
| 502 |
self.session_data = []
|
| 503 |
images_saved = 0
|
| 504 |
|
| 505 |
-
print(f"📊 Assemblage de {total_questions} résultats...")
|
| 506 |
|
| 507 |
for i in range(total_questions):
|
| 508 |
if i in self.results_cache:
|
| 509 |
row_data = self.results_cache[i]
|
| 510 |
-
print(f" ✅ Résultat {i} du cache parallèle")
|
| 511 |
else:
|
| 512 |
-
print(f" 🔄 Traitement fallback pour résultat {i}...")
|
| 513 |
if i < len(self.operations_history):
|
| 514 |
row_data = create_result_row_with_images(i, self.user_images[i], self.expected_answers[i], self.operations_history[i])
|
| 515 |
else:
|
|
@@ -526,15 +471,15 @@ class MathGame:
|
|
| 526 |
if row_data['is_correct']:
|
| 527 |
correct_answers += 1
|
| 528 |
|
| 529 |
-
# Structure pour dataset avec
|
| 530 |
a, b, operation, correct_result = self.operations_history[i] if i < len(self.operations_history) else (0, 0, "×", 0)
|
| 531 |
|
| 532 |
try:
|
| 533 |
ocr_info_data = get_ocr_model_info()
|
| 534 |
-
print(f"🔍 Debug OCR info: {ocr_info_data}")
|
| 535 |
except Exception as e:
|
| 536 |
print(f"❌ Erreur get_ocr_model_info: {e}")
|
| 537 |
-
ocr_info_data = {"model_name": "
|
| 538 |
|
| 539 |
entry = {
|
| 540 |
"session_id": session_id,
|
|
@@ -547,18 +492,18 @@ class MathGame:
|
|
| 547 |
"operand_b": b,
|
| 548 |
"operation": operation,
|
| 549 |
"correct_answer": self.expected_answers[i] if i < len(self.expected_answers) else 0,
|
| 550 |
-
"ocr_model": ocr_info_data.get("model_name", "
|
| 551 |
-
"ocr_device": ocr_info_data.get("device", "
|
| 552 |
"user_answer_ocr": row_data['recognized'],
|
| 553 |
"user_answer_parsed": row_data['recognized_num'],
|
| 554 |
"is_correct": row_data['is_correct'],
|
| 555 |
"total_questions": total_questions,
|
| 556 |
-
"app_version": "3.
|
| 557 |
}
|
| 558 |
|
| 559 |
-
print(f"🔍 Debug entry OCR
|
| 560 |
|
| 561 |
-
#
|
| 562 |
if row_data['dataset_image_data']:
|
| 563 |
entry["handwriting_image"] = row_data['dataset_image_data']["handwriting_image"] # Image PIL native
|
| 564 |
entry["image_width"] = int(row_data['dataset_image_data']["width"])
|
|
@@ -575,7 +520,7 @@ class MathGame:
|
|
| 575 |
for entry in self.session_data:
|
| 576 |
entry["session_accuracy"] = accuracy
|
| 577 |
|
| 578 |
-
# Nettoyage mémoire
|
| 579 |
for img in self.user_images:
|
| 580 |
if hasattr(img, 'close'):
|
| 581 |
try:
|
|
@@ -597,7 +542,7 @@ class MathGame:
|
|
| 597 |
<th style="padding: 8px;">B</th>
|
| 598 |
<th style="padding: 8px;">Réponse</th>
|
| 599 |
<th style="padding: 8px;">Votre dessin</th>
|
| 600 |
-
<th style="padding: 8px;">OCR</th>
|
| 601 |
<th style="padding: 8px;">Statut</th>
|
| 602 |
</tr>
|
| 603 |
</thead>
|
|
@@ -619,10 +564,11 @@ class MathGame:
|
|
| 619 |
if export_info["can_export"]:
|
| 620 |
export_section = f"""
|
| 621 |
<div style="margin-top: 20px; padding: 15px; background-color: #e8f5e8; border-radius: 8px;">
|
| 622 |
-
<h3 style="color: #2e7d32;">📊 Résumé de la série</h3>
|
| 623 |
<p style="color: #2e7d32;">
|
| 624 |
✅ {total_questions} réponses • 📊 {accuracy:.1f}% de précision<br>
|
| 625 |
🖼️ {images_saved} images natives sauvegardées<br>
|
|
|
|
| 626 |
⚙️ Configuration: {config_display}
|
| 627 |
</p>
|
| 628 |
</div>
|
|
@@ -653,7 +599,7 @@ class MathGame:
|
|
| 653 |
</div>
|
| 654 |
</div>
|
| 655 |
</div>
|
| 656 |
-
<h2 style="color: #4a90e2;">📊 Détail des Réponses</h2>
|
| 657 |
{table_html}
|
| 658 |
{export_section}
|
| 659 |
</div>
|
|
@@ -662,7 +608,7 @@ class MathGame:
|
|
| 662 |
return (
|
| 663 |
"""<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">🏁 C'est fini !</div>""",
|
| 664 |
create_white_canvas(),
|
| 665 |
-
f"✨ Session {config_display} terminée !",
|
| 666 |
"⏱️ Temps écoulé !",
|
| 667 |
gr.update(interactive=True),
|
| 668 |
gr.update(interactive=False),
|
|
@@ -671,9 +617,9 @@ class MathGame:
|
|
| 671 |
|
| 672 |
|
| 673 |
def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None) -> str:
|
| 674 |
-
"""Export vers le
|
| 675 |
if dataset_name is None:
|
| 676 |
-
dataset_name = DATASET_NAME
|
| 677 |
|
| 678 |
if not DATASET_AVAILABLE:
|
| 679 |
return "❌ Modules dataset non disponibles"
|
|
@@ -683,41 +629,40 @@ def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None)
|
|
| 683 |
return "❌ Token HuggingFace manquant (HF_TOKEN ou tk_calcul_ocr)"
|
| 684 |
|
| 685 |
try:
|
| 686 |
-
print(f"\n🚀 === EXPORT VERS DATASET CALCUL OCR
|
| 687 |
print(f"📊 Dataset: {dataset_name}")
|
| 688 |
|
| 689 |
-
# Filtrer les entrées avec images et ajouter les infos OCR
|
| 690 |
clean_entries = []
|
| 691 |
|
| 692 |
-
# Récupérer
|
| 693 |
try:
|
| 694 |
global_ocr_info = get_ocr_model_info()
|
| 695 |
-
print(f"🔍 Infos OCR globales: {global_ocr_info}")
|
| 696 |
except Exception as e:
|
| 697 |
-
print(f"❌ Erreur infos OCR
|
| 698 |
-
global_ocr_info = {"model_name": "
|
| 699 |
|
| 700 |
for entry in session_data:
|
| 701 |
if entry.get('has_image', False):
|
| 702 |
-
# Ajouter explicitement les champs OCR
|
| 703 |
entry_with_ocr = entry.copy()
|
| 704 |
-
entry_with_ocr["ocr_model"] = global_ocr_info.get("model_name", "
|
| 705 |
-
entry_with_ocr["ocr_device"] = global_ocr_info.get("device", "
|
| 706 |
|
| 707 |
-
print(f"🔍 Entry avec OCR: ocr_model={entry_with_ocr['ocr_model']}, ocr_device={entry_with_ocr['ocr_device']}")
|
| 708 |
print(f"🖼️ Image type: {type(entry_with_ocr.get('handwriting_image', 'None'))}")
|
| 709 |
clean_entries.append(entry_with_ocr)
|
| 710 |
|
| 711 |
-
# Créer un dataset de test avec structure forcée
|
| 712 |
if len(clean_entries) == 0:
|
| 713 |
return "❌ Aucune entrée avec image à exporter"
|
| 714 |
|
| 715 |
# Vérifier la structure de la première entrée
|
| 716 |
sample_entry = clean_entries[0]
|
| 717 |
-
print(f"🔍 Structure première entrée: {list(sample_entry.keys())}")
|
| 718 |
-
print(f"🔍 OCR dans entrée: ocr_model={sample_entry.get('ocr_model', 'MISSING')}, ocr_device={sample_entry.get('ocr_device', 'MISSING')}")
|
| 719 |
|
| 720 |
-
# Charger dataset existant et combiner
|
| 721 |
try:
|
| 722 |
existing_dataset = load_dataset(dataset_name, split="train")
|
| 723 |
existing_data = existing_dataset.to_list()
|
|
@@ -734,14 +679,14 @@ def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None)
|
|
| 734 |
clean_dataset = Dataset.from_list(clean_entries)
|
| 735 |
print(f"📊 Nouveau dataset créé avec {len(clean_entries)} entrées")
|
| 736 |
|
| 737 |
-
#
|
| 738 |
try:
|
| 739 |
clean_dataset = clean_dataset.cast_column("handwriting_image", DatasetImage())
|
| 740 |
print("✅ Colonne handwriting_image convertie au type Image natif")
|
| 741 |
except Exception as e:
|
| 742 |
print(f"⚠️ Impossible de convertir au type Image: {e}")
|
| 743 |
|
| 744 |
-
print(f"✅ Dataset créé - Features:")
|
| 745 |
for feature_name in clean_dataset.features:
|
| 746 |
print(f" - {feature_name}: {clean_dataset.features[feature_name]}")
|
| 747 |
|
|
@@ -754,20 +699,21 @@ def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None)
|
|
| 754 |
operations_summary = ", ".join([f"{op}: {count}" for op, count in operations_count.items()])
|
| 755 |
|
| 756 |
# Push vers HuggingFace
|
| 757 |
-
print(f"📤 Push vers {dataset_name}...")
|
| 758 |
clean_dataset.push_to_hub(
|
| 759 |
dataset_name,
|
| 760 |
private=False,
|
| 761 |
token=hf_token,
|
| 762 |
-
commit_message=f"Add {len(clean_entries)} native image samples for math OCR ({operations_summary})"
|
| 763 |
)
|
| 764 |
|
| 765 |
cleanup_memory()
|
| 766 |
|
| 767 |
-
return f"""### ✅ Session ajoutée au dataset avec succès !
|
| 768 |
|
| 769 |
📊 **Dataset:** {dataset_name}
|
| 770 |
-
🖼️ **Images natives:** {len(clean_entries)}
|
|
|
|
| 771 |
🔢 **Opérations:** {operations_summary}
|
| 772 |
📈 **Total:** {len(clean_dataset)}
|
| 773 |
|
|
@@ -776,7 +722,7 @@ def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None)
|
|
| 776 |
|
| 777 |
|
| 778 |
except Exception as e:
|
| 779 |
-
print(f"❌ ERREUR: {e}")
|
| 780 |
import traceback
|
| 781 |
traceback.print_exc()
|
| 782 |
-
return f"❌ Erreur: {str(e)}"
|
|
|
|
| 1 |
# ==========================================
|
| 2 |
+
# game_engine.py - Calcul OCR v3.0 GPU simplifié
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
+
Moteur de jeu mathématique optimisé pour ZeroGPU HuggingFace Spaces
|
|
|
|
| 7 |
"""
|
| 8 |
|
| 9 |
import random
|
|
|
|
| 19 |
import queue
|
| 20 |
from typing import Dict, Tuple, Optional
|
| 21 |
|
| 22 |
+
# Import GPU uniquement
|
| 23 |
+
from image_processing_gpu import (
|
| 24 |
+
recognize_number_fast_with_image,
|
| 25 |
+
create_thumbnail_fast,
|
| 26 |
+
create_white_canvas,
|
| 27 |
+
cleanup_memory,
|
| 28 |
+
log_memory_usage,
|
| 29 |
+
get_ocr_model_info
|
| 30 |
+
)
|
| 31 |
|
| 32 |
+
print("✅ Game Engine: Mode GPU TrOCR")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
+
# Récupérer les infos du modèle OCR
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
try:
|
| 36 |
ocr_info = get_ocr_model_info()
|
| 37 |
print(f"🎯 OCR sélectionné: {ocr_info['model_name']} sur {ocr_info['device']}")
|
| 38 |
except Exception as e:
|
| 39 |
print(f"⚠️ Impossible de récupérer les infos OCR: {e}")
|
| 40 |
+
ocr_info = {"model_name": "TrOCR", "device": "ZeroGPU"}
|
| 41 |
|
| 42 |
+
# Imports dataset
|
| 43 |
try:
|
| 44 |
from datasets import Dataset, Image as DatasetImage, load_dataset
|
| 45 |
DATASET_AVAILABLE = True
|
|
|
|
| 48 |
DATASET_AVAILABLE = False
|
| 49 |
print(f"⚠️ Modules dataset non disponibles: {e}")
|
| 50 |
|
| 51 |
+
# Dataset name
|
| 52 |
DATASET_NAME = "hoololi/calcul_ocr_dataset_with_images_v2"
|
| 53 |
|
| 54 |
# Configuration des difficultés par opération
|
|
|
|
| 115 |
|
| 116 |
|
| 117 |
class MathGame:
|
| 118 |
+
"""Moteur de jeu mathématique avec traitement parallèle GPU"""
|
| 119 |
|
| 120 |
def __init__(self):
|
| 121 |
self.is_running = False
|
|
|
|
| 167 |
self.processing_active = True
|
| 168 |
self.worker_thread = threading.Thread(target=self._process_images_worker, daemon=True)
|
| 169 |
self.worker_thread.start()
|
| 170 |
+
print("🔄 Thread de traitement GPU parallèle démarré")
|
| 171 |
|
| 172 |
def _stop_background_processing(self) -> None:
|
| 173 |
"""Arrête le thread de traitement"""
|
| 174 |
self.processing_active = False
|
| 175 |
if self.worker_thread and self.worker_thread.is_alive():
|
| 176 |
+
print("⏹️ Arrêt du thread de traitement GPU parallèle")
|
| 177 |
|
| 178 |
def _process_images_worker(self) -> None:
|
| 179 |
+
"""Worker thread qui traite les images en arrière-plan avec GPU"""
|
| 180 |
+
print("🚀 Worker thread GPU démarré")
|
| 181 |
while self.processing_active:
|
| 182 |
try:
|
| 183 |
if not self.processing_queue.empty():
|
| 184 |
question_num, image, expected, operation_data = self.processing_queue.get(timeout=1)
|
| 185 |
+
print(f"🔄 Traitement GPU parallèle image {question_num}...")
|
| 186 |
|
| 187 |
start_time = time.time()
|
| 188 |
result_data = create_result_row_with_images(question_num, image, expected, operation_data)
|
|
|
|
| 190 |
|
| 191 |
# Stocker le résultat
|
| 192 |
self.results_cache[question_num] = result_data
|
| 193 |
+
print(f"✅ Image {question_num} traitée en {processing_time:.1f}s (GPU parallèle)")
|
| 194 |
|
| 195 |
else:
|
| 196 |
time.sleep(0.1)
|
|
|
|
| 198 |
except queue.Empty:
|
| 199 |
continue
|
| 200 |
except Exception as e:
|
| 201 |
+
print(f"❌ Erreur traitement GPU parallèle: {e}")
|
| 202 |
|
| 203 |
+
print("🛑 Worker thread GPU terminé")
|
| 204 |
|
| 205 |
def _add_image_to_processing_queue(self, question_num: int, image: dict | np.ndarray | Image.Image,
|
| 206 |
expected: int, operation_data: tuple) -> None:
|
| 207 |
+
"""Ajoute une image à la queue de traitement GPU"""
|
| 208 |
if image is not None:
|
| 209 |
self.processing_queue.put((question_num, image, expected, operation_data))
|
| 210 |
+
print(f"📝 Image {question_num} ajoutée à la queue de traitement GPU")
|
| 211 |
|
| 212 |
def generate_multiplication(self, difficulty: str) -> tuple[str, int]:
|
| 213 |
"""Génère une multiplication"""
|
|
|
|
| 303 |
self.export_timestamp = None
|
| 304 |
self.export_result = None
|
| 305 |
|
| 306 |
+
# Démarrer le traitement parallèle GPU
|
| 307 |
self._start_background_processing()
|
| 308 |
|
| 309 |
gc.collect()
|
|
|
|
| 351 |
return self.end_game(image_data)
|
| 352 |
|
| 353 |
if image_data is not None:
|
| 354 |
+
# Ajouter l'image à la liste ET au traitement parallèle GPU
|
| 355 |
self.user_images.append(image_data)
|
| 356 |
self.expected_answers.append(self.correct_answer)
|
| 357 |
|
|
|
|
| 360 |
a, op, b = int(parts[0]), parts[1], int(parts[2])
|
| 361 |
current_operation_data = (a, b, op, self.correct_answer)
|
| 362 |
|
| 363 |
+
# Lancer le traitement GPU en parallèle de l'image qu'on vient de recevoir
|
| 364 |
self._add_image_to_processing_queue(self.question_count, image_data, self.correct_answer, current_operation_data)
|
| 365 |
|
| 366 |
self.question_count += 1
|
|
|
|
| 401 |
|
| 402 |
self.is_running = False
|
| 403 |
|
| 404 |
+
# Arrêter le traitement parallèle GPU
|
| 405 |
self._stop_background_processing()
|
| 406 |
|
| 407 |
+
print("🏁 Fin de jeu - Assemblage des résultats GPU...")
|
| 408 |
|
| 409 |
if final_image is not None:
|
| 410 |
self.user_images.append(final_image)
|
|
|
|
| 415 |
a, op, b = int(parts[0]), parts[1], int(parts[2])
|
| 416 |
final_operation_data = (a, b, op, self.correct_answer)
|
| 417 |
|
| 418 |
+
# Traiter la dernière image immédiatement avec GPU
|
| 419 |
+
print(f"🔄 Traitement GPU final de l'image {self.question_count}...")
|
| 420 |
final_result = create_result_row_with_images(self.question_count, final_image, self.correct_answer, final_operation_data)
|
| 421 |
self.results_cache[self.question_count] = final_result
|
| 422 |
|
|
|
|
| 424 |
if len(self.operations_history) < len(self.user_images):
|
| 425 |
self.operations_history.append((a, b, op, self.correct_answer))
|
| 426 |
|
| 427 |
+
# Attendre que toutes les images soient traitées par GPU
|
| 428 |
max_wait = 10
|
| 429 |
wait_start = time.time()
|
| 430 |
expected_results = len(self.user_images)
|
| 431 |
|
| 432 |
+
print(f"⏳ Attente de {expected_results} résultats GPU...")
|
| 433 |
while len(self.results_cache) < expected_results and (time.time() - wait_start) < max_wait:
|
| 434 |
time.sleep(0.1)
|
| 435 |
|
| 436 |
results_ready = len(self.results_cache)
|
| 437 |
+
print(f"✅ {results_ready}/{expected_results} résultats GPU prêts")
|
| 438 |
|
| 439 |
# Assembler les résultats dans l'ordre
|
| 440 |
correct_answers = 0
|
|
|
|
| 447 |
self.session_data = []
|
| 448 |
images_saved = 0
|
| 449 |
|
| 450 |
+
print(f"📊 Assemblage de {total_questions} résultats GPU...")
|
| 451 |
|
| 452 |
for i in range(total_questions):
|
| 453 |
if i in self.results_cache:
|
| 454 |
row_data = self.results_cache[i]
|
| 455 |
+
print(f" ✅ Résultat {i} du cache GPU parallèle")
|
| 456 |
else:
|
| 457 |
+
print(f" 🔄 Traitement GPU fallback pour résultat {i}...")
|
| 458 |
if i < len(self.operations_history):
|
| 459 |
row_data = create_result_row_with_images(i, self.user_images[i], self.expected_answers[i], self.operations_history[i])
|
| 460 |
else:
|
|
|
|
| 471 |
if row_data['is_correct']:
|
| 472 |
correct_answers += 1
|
| 473 |
|
| 474 |
+
# Structure pour dataset avec info OCR GPU
|
| 475 |
a, b, operation, correct_result = self.operations_history[i] if i < len(self.operations_history) else (0, 0, "×", 0)
|
| 476 |
|
| 477 |
try:
|
| 478 |
ocr_info_data = get_ocr_model_info()
|
| 479 |
+
print(f"🔍 Debug OCR info GPU: {ocr_info_data}")
|
| 480 |
except Exception as e:
|
| 481 |
print(f"❌ Erreur get_ocr_model_info: {e}")
|
| 482 |
+
ocr_info_data = {"model_name": "TrOCR", "device": "ZeroGPU"}
|
| 483 |
|
| 484 |
entry = {
|
| 485 |
"session_id": session_id,
|
|
|
|
| 492 |
"operand_b": b,
|
| 493 |
"operation": operation,
|
| 494 |
"correct_answer": self.expected_answers[i] if i < len(self.expected_answers) else 0,
|
| 495 |
+
"ocr_model": ocr_info_data.get("model_name", "TrOCR"),
|
| 496 |
+
"ocr_device": ocr_info_data.get("device", "ZeroGPU"),
|
| 497 |
"user_answer_ocr": row_data['recognized'],
|
| 498 |
"user_answer_parsed": row_data['recognized_num'],
|
| 499 |
"is_correct": row_data['is_correct'],
|
| 500 |
"total_questions": total_questions,
|
| 501 |
+
"app_version": "3.0_calcul_ocr_gpu_simplified"
|
| 502 |
}
|
| 503 |
|
| 504 |
+
print(f"🔍 Debug entry OCR GPU: ocr_model={entry['ocr_model']}, ocr_device={entry['ocr_device']}")
|
| 505 |
|
| 506 |
+
# Image PIL native pour dataset
|
| 507 |
if row_data['dataset_image_data']:
|
| 508 |
entry["handwriting_image"] = row_data['dataset_image_data']["handwriting_image"] # Image PIL native
|
| 509 |
entry["image_width"] = int(row_data['dataset_image_data']["width"])
|
|
|
|
| 520 |
for entry in self.session_data:
|
| 521 |
entry["session_accuracy"] = accuracy
|
| 522 |
|
| 523 |
+
# Nettoyage mémoire GPU
|
| 524 |
for img in self.user_images:
|
| 525 |
if hasattr(img, 'close'):
|
| 526 |
try:
|
|
|
|
| 542 |
<th style="padding: 8px;">B</th>
|
| 543 |
<th style="padding: 8px;">Réponse</th>
|
| 544 |
<th style="padding: 8px;">Votre dessin</th>
|
| 545 |
+
<th style="padding: 8px;">OCR GPU</th>
|
| 546 |
<th style="padding: 8px;">Statut</th>
|
| 547 |
</tr>
|
| 548 |
</thead>
|
|
|
|
| 564 |
if export_info["can_export"]:
|
| 565 |
export_section = f"""
|
| 566 |
<div style="margin-top: 20px; padding: 15px; background-color: #e8f5e8; border-radius: 8px;">
|
| 567 |
+
<h3 style="color: #2e7d32;">📊 Résumé de la série (GPU)</h3>
|
| 568 |
<p style="color: #2e7d32;">
|
| 569 |
✅ {total_questions} réponses • 📊 {accuracy:.1f}% de précision<br>
|
| 570 |
🖼️ {images_saved} images natives sauvegardées<br>
|
| 571 |
+
🎯 OCR: TrOCR ZeroGPU<br>
|
| 572 |
⚙️ Configuration: {config_display}
|
| 573 |
</p>
|
| 574 |
</div>
|
|
|
|
| 599 |
</div>
|
| 600 |
</div>
|
| 601 |
</div>
|
| 602 |
+
<h2 style="color: #4a90e2;">📊 Détail des Réponses (TrOCR GPU)</h2>
|
| 603 |
{table_html}
|
| 604 |
{export_section}
|
| 605 |
</div>
|
|
|
|
| 608 |
return (
|
| 609 |
"""<div style="font-size: 3em; font-weight: bold; text-align: center; padding: 20px; background: linear-gradient(45deg, #667eea 0%, #764ba2 100%); color: white; border-radius: 10px;">🏁 C'est fini !</div>""",
|
| 610 |
create_white_canvas(),
|
| 611 |
+
f"✨ Session GPU {config_display} terminée !",
|
| 612 |
"⏱️ Temps écoulé !",
|
| 613 |
gr.update(interactive=True),
|
| 614 |
gr.update(interactive=False),
|
|
|
|
| 617 |
|
| 618 |
|
| 619 |
def export_to_clean_dataset(session_data: list[dict], dataset_name: str = None) -> str:
|
| 620 |
+
"""Export vers le dataset avec images natives optimisé GPU"""
|
| 621 |
if dataset_name is None:
|
| 622 |
+
dataset_name = DATASET_NAME
|
| 623 |
|
| 624 |
if not DATASET_AVAILABLE:
|
| 625 |
return "❌ Modules dataset non disponibles"
|
|
|
|
| 629 |
return "❌ Token HuggingFace manquant (HF_TOKEN ou tk_calcul_ocr)"
|
| 630 |
|
| 631 |
try:
|
| 632 |
+
print(f"\n🚀 === EXPORT VERS DATASET CALCUL OCR GPU ===")
|
| 633 |
print(f"📊 Dataset: {dataset_name}")
|
| 634 |
|
| 635 |
+
# Filtrer les entrées avec images et ajouter les infos OCR GPU
|
| 636 |
clean_entries = []
|
| 637 |
|
| 638 |
+
# Récupérer les infos OCR GPU pour toute la session
|
| 639 |
try:
|
| 640 |
global_ocr_info = get_ocr_model_info()
|
| 641 |
+
print(f"🔍 Infos OCR GPU globales: {global_ocr_info}")
|
| 642 |
except Exception as e:
|
| 643 |
+
print(f"❌ Erreur infos OCR GPU: {e}")
|
| 644 |
+
global_ocr_info = {"model_name": "TrOCR", "device": "ZeroGPU"}
|
| 645 |
|
| 646 |
for entry in session_data:
|
| 647 |
if entry.get('has_image', False):
|
| 648 |
+
# Ajouter explicitement les champs OCR GPU
|
| 649 |
entry_with_ocr = entry.copy()
|
| 650 |
+
entry_with_ocr["ocr_model"] = global_ocr_info.get("model_name", "TrOCR")
|
| 651 |
+
entry_with_ocr["ocr_device"] = global_ocr_info.get("device", "ZeroGPU")
|
| 652 |
|
| 653 |
+
print(f"🔍 Entry GPU avec OCR: ocr_model={entry_with_ocr['ocr_model']}, ocr_device={entry_with_ocr['ocr_device']}")
|
| 654 |
print(f"🖼️ Image type: {type(entry_with_ocr.get('handwriting_image', 'None'))}")
|
| 655 |
clean_entries.append(entry_with_ocr)
|
| 656 |
|
|
|
|
| 657 |
if len(clean_entries) == 0:
|
| 658 |
return "❌ Aucune entrée avec image à exporter"
|
| 659 |
|
| 660 |
# Vérifier la structure de la première entrée
|
| 661 |
sample_entry = clean_entries[0]
|
| 662 |
+
print(f"🔍 Structure première entrée GPU: {list(sample_entry.keys())}")
|
| 663 |
+
print(f"🔍 OCR GPU dans entrée: ocr_model={sample_entry.get('ocr_model', 'MISSING')}, ocr_device={sample_entry.get('ocr_device', 'MISSING')}")
|
| 664 |
|
| 665 |
+
# Charger dataset existant et combiner
|
| 666 |
try:
|
| 667 |
existing_dataset = load_dataset(dataset_name, split="train")
|
| 668 |
existing_data = existing_dataset.to_list()
|
|
|
|
| 679 |
clean_dataset = Dataset.from_list(clean_entries)
|
| 680 |
print(f"📊 Nouveau dataset créé avec {len(clean_entries)} entrées")
|
| 681 |
|
| 682 |
+
# Spécifier que la colonne handwriting_image est de type Image
|
| 683 |
try:
|
| 684 |
clean_dataset = clean_dataset.cast_column("handwriting_image", DatasetImage())
|
| 685 |
print("✅ Colonne handwriting_image convertie au type Image natif")
|
| 686 |
except Exception as e:
|
| 687 |
print(f"⚠️ Impossible de convertir au type Image: {e}")
|
| 688 |
|
| 689 |
+
print(f"✅ Dataset GPU créé - Features:")
|
| 690 |
for feature_name in clean_dataset.features:
|
| 691 |
print(f" - {feature_name}: {clean_dataset.features[feature_name]}")
|
| 692 |
|
|
|
|
| 699 |
operations_summary = ", ".join([f"{op}: {count}" for op, count in operations_count.items()])
|
| 700 |
|
| 701 |
# Push vers HuggingFace
|
| 702 |
+
print(f"📤 Push GPU vers {dataset_name}...")
|
| 703 |
clean_dataset.push_to_hub(
|
| 704 |
dataset_name,
|
| 705 |
private=False,
|
| 706 |
token=hf_token,
|
| 707 |
+
commit_message=f"Add {len(clean_entries)} GPU TrOCR native image samples for math OCR ({operations_summary})"
|
| 708 |
)
|
| 709 |
|
| 710 |
cleanup_memory()
|
| 711 |
|
| 712 |
+
return f"""### ✅ Session GPU ajoutée au dataset avec succès !
|
| 713 |
|
| 714 |
📊 **Dataset:** {dataset_name}
|
| 715 |
+
🖼️ **Images natives (GPU):** {len(clean_entries)}
|
| 716 |
+
🎯 **OCR:** TrOCR ZeroGPU
|
| 717 |
🔢 **Opérations:** {operations_summary}
|
| 718 |
📈 **Total:** {len(clean_dataset)}
|
| 719 |
|
|
|
|
| 722 |
|
| 723 |
|
| 724 |
except Exception as e:
|
| 725 |
+
print(f"❌ ERREUR GPU: {e}")
|
| 726 |
import traceback
|
| 727 |
traceback.print_exc()
|
| 728 |
+
return f"❌ Erreur GPU: {str(e)}"
|
image_processing_gpu.py
CHANGED
|
@@ -1,37 +1,15 @@
|
|
| 1 |
# ==========================================
|
| 2 |
-
# image_processing_gpu.py - Version ZeroGPU
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
-
Module de traitement d'images GPU-optimisé pour
|
| 7 |
-
Compatible ZeroGPU HuggingFace Spaces
|
| 8 |
-
VERSION MODIFIÉE: Format Image natif au lieu de Base64
|
| 9 |
"""
|
| 10 |
|
| 11 |
import time
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
import spaces
|
| 16 |
-
print("✅ Import spaces réussi dans image_processing_gpu")
|
| 17 |
-
SPACES_AVAILABLE = True
|
| 18 |
-
except ImportError as e:
|
| 19 |
-
print(f"❌ Import spaces échoué: {e}")
|
| 20 |
-
# Créer un mock si spaces n'est pas disponible
|
| 21 |
-
class MockSpaces:
|
| 22 |
-
@staticmethod
|
| 23 |
-
def GPU(func):
|
| 24 |
-
print(f"MockSpaces.GPU décorateur appliqué à {func.__name__}")
|
| 25 |
-
return func
|
| 26 |
-
spaces = MockSpaces()
|
| 27 |
-
SPACES_AVAILABLE = False
|
| 28 |
-
|
| 29 |
-
try:
|
| 30 |
-
import torch
|
| 31 |
-
TORCH_AVAILABLE = True
|
| 32 |
-
except ImportError:
|
| 33 |
-
print("❌ Torch non disponible")
|
| 34 |
-
TORCH_AVAILABLE = False
|
| 35 |
|
| 36 |
from utils import (
|
| 37 |
optimize_image_for_ocr,
|
|
@@ -49,17 +27,11 @@ model = None
|
|
| 49 |
OCR_MODEL_NAME = "TrOCR-base-handwritten"
|
| 50 |
|
| 51 |
def init_ocr_model() -> bool:
|
| 52 |
-
"""Initialise TrOCR
|
| 53 |
global processor, model
|
| 54 |
|
| 55 |
try:
|
| 56 |
-
print("🔄 Chargement TrOCR (ZeroGPU
|
| 57 |
-
|
| 58 |
-
if not TORCH_AVAILABLE:
|
| 59 |
-
print("❌ Torch non disponible, impossible de charger TrOCR")
|
| 60 |
-
return False
|
| 61 |
-
|
| 62 |
-
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
|
| 63 |
|
| 64 |
processor = TrOCRProcessor.from_pretrained('microsoft/trocr-base-handwritten')
|
| 65 |
model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-base-handwritten')
|
|
@@ -83,9 +55,9 @@ def init_ocr_model() -> bool:
|
|
| 83 |
|
| 84 |
def get_ocr_model_info() -> dict:
|
| 85 |
"""Retourne les informations du modèle OCR utilisé"""
|
| 86 |
-
if
|
| 87 |
device = "ZeroGPU"
|
| 88 |
-
gpu_name = torch.cuda.get_device_name()
|
| 89 |
else:
|
| 90 |
device = "CPU"
|
| 91 |
gpu_name = "N/A"
|
|
@@ -99,10 +71,10 @@ def get_ocr_model_info() -> dict:
|
|
| 99 |
"version": "microsoft/trocr-base-handwritten"
|
| 100 |
}
|
| 101 |
|
| 102 |
-
@spaces.GPU
|
| 103 |
def recognize_number_fast_with_image(image_dict, debug: bool = False) -> tuple[str, any, dict | None]:
|
| 104 |
"""
|
| 105 |
-
OCR avec TrOCR
|
| 106 |
"""
|
| 107 |
if image_dict is None:
|
| 108 |
if debug:
|
|
@@ -152,7 +124,7 @@ def recognize_number_fast_with_image(image_dict, debug: bool = False) -> tuple[s
|
|
| 152 |
result = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
| 153 |
final_result = validate_ocr_result(result, max_length=4)
|
| 154 |
|
| 155 |
-
# Préparer pour dataset
|
| 156 |
dataset_image_data = prepare_image_for_dataset(optimized_image)
|
| 157 |
|
| 158 |
if debug:
|
|
|
|
| 1 |
# ==========================================
|
| 2 |
+
# image_processing_gpu.py - Version ZeroGPU simplifiée
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
+
Module de traitement d'images GPU-optimisé pour ZeroGPU HuggingFace Spaces
|
|
|
|
|
|
|
| 7 |
"""
|
| 8 |
|
| 9 |
import time
|
| 10 |
+
import torch
|
| 11 |
+
import spaces
|
| 12 |
+
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
from utils import (
|
| 15 |
optimize_image_for_ocr,
|
|
|
|
| 27 |
OCR_MODEL_NAME = "TrOCR-base-handwritten"
|
| 28 |
|
| 29 |
def init_ocr_model() -> bool:
|
| 30 |
+
"""Initialise TrOCR pour ZeroGPU"""
|
| 31 |
global processor, model
|
| 32 |
|
| 33 |
try:
|
| 34 |
+
print("🔄 Chargement TrOCR (ZeroGPU)...")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
processor = TrOCRProcessor.from_pretrained('microsoft/trocr-base-handwritten')
|
| 37 |
model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-base-handwritten')
|
|
|
|
| 55 |
|
| 56 |
def get_ocr_model_info() -> dict:
|
| 57 |
"""Retourne les informations du modèle OCR utilisé"""
|
| 58 |
+
if torch.cuda.is_available():
|
| 59 |
device = "ZeroGPU"
|
| 60 |
+
gpu_name = torch.cuda.get_device_name()
|
| 61 |
else:
|
| 62 |
device = "CPU"
|
| 63 |
gpu_name = "N/A"
|
|
|
|
| 71 |
"version": "microsoft/trocr-base-handwritten"
|
| 72 |
}
|
| 73 |
|
| 74 |
+
@spaces.GPU
|
| 75 |
def recognize_number_fast_with_image(image_dict, debug: bool = False) -> tuple[str, any, dict | None]:
|
| 76 |
"""
|
| 77 |
+
OCR avec TrOCR ZeroGPU - Version simplifiée
|
| 78 |
"""
|
| 79 |
if image_dict is None:
|
| 80 |
if debug:
|
|
|
|
| 124 |
result = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
| 125 |
final_result = validate_ocr_result(result, max_length=4)
|
| 126 |
|
| 127 |
+
# Préparer pour dataset
|
| 128 |
dataset_image_data = prepare_image_for_dataset(optimized_image)
|
| 129 |
|
| 130 |
if debug:
|
requirements.txt
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
# Requirements
|
| 2 |
gradio>=4.0.0
|
| 3 |
pillow>=9.0.0
|
| 4 |
numpy>=1.21.0
|
|
@@ -7,11 +7,8 @@ huggingface_hub>=0.16.0
|
|
| 7 |
pandas>=1.5.0
|
| 8 |
psutil>=5.8.0
|
| 9 |
|
| 10 |
-
#
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
torch>=2.0.0; sys_platform != "emscripten"
|
| 16 |
-
torchvision>=0.15.0; sys_platform != "emscripten"
|
| 17 |
-
transformers>=4.30.0; sys_platform != "emscripten"
|
|
|
|
| 1 |
+
# Requirements simplifiés pour GPU/ZeroGPU uniquement
|
| 2 |
gradio>=4.0.0
|
| 3 |
pillow>=9.0.0
|
| 4 |
numpy>=1.21.0
|
|
|
|
| 7 |
pandas>=1.5.0
|
| 8 |
psutil>=5.8.0
|
| 9 |
|
| 10 |
+
# GPU/TrOCR pour ZeroGPU HuggingFace Spaces
|
| 11 |
+
torch>=2.0.0
|
| 12 |
+
torchvision>=0.15.0
|
| 13 |
+
transformers>=4.30.0
|
| 14 |
+
spaces
|
|
|
|
|
|
|
|
|
utils.py
CHANGED
|
@@ -1,17 +1,15 @@
|
|
| 1 |
# ==========================================
|
| 2 |
-
# utils.py - Fonctions communes
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
-
Utilitaires partagés pour le traitement d'images OCR
|
| 7 |
-
Fonctions communes aux versions CPU et GPU - Format Image natif
|
| 8 |
"""
|
| 9 |
|
| 10 |
from PIL import Image, ImageEnhance
|
| 11 |
import numpy as np
|
| 12 |
import gc
|
| 13 |
import os
|
| 14 |
-
import time
|
| 15 |
|
| 16 |
def create_white_canvas(width: int = 300, height: int = 300) -> Image.Image:
|
| 17 |
"""Crée un canvas blanc pour le dessin de calculs"""
|
|
@@ -28,12 +26,18 @@ def log_memory_usage(context: str = "") -> None:
|
|
| 28 |
pass
|
| 29 |
|
| 30 |
def cleanup_memory() -> None:
|
| 31 |
-
"""Force le nettoyage mémoire"""
|
| 32 |
gc.collect()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
def optimize_image_for_ocr(image_dict: dict | np.ndarray | Image.Image | None, max_size: int = 300) -> Image.Image | None:
|
| 35 |
"""
|
| 36 |
-
Optimisation image
|
| 37 |
|
| 38 |
Args:
|
| 39 |
image_dict: Image d'entrée (format Gradio, numpy ou PIL)
|
|
@@ -121,14 +125,13 @@ def create_thumbnail_fast(optimized_image: Image.Image | None, size: tuple[int,
|
|
| 121 |
size: Taille de la miniature (largeur, hauteur)
|
| 122 |
|
| 123 |
Returns:
|
| 124 |
-
HTML img tag avec image base64 ou icône par défaut
|
| 125 |
"""
|
| 126 |
try:
|
| 127 |
if optimized_image is None:
|
| 128 |
return "📝"
|
| 129 |
|
| 130 |
# Pour l'affichage dans l'interface, on garde le base64 temporairement
|
| 131 |
-
# mais seulement pour les miniatures d'affichage
|
| 132 |
import base64
|
| 133 |
from io import BytesIO
|
| 134 |
|
|
@@ -171,41 +174,4 @@ def validate_ocr_result(raw_result: str, max_length: int = 4) -> str:
|
|
| 171 |
# Si trop long, prendre les premiers chiffres
|
| 172 |
return cleaned_result[:max_length]
|
| 173 |
else:
|
| 174 |
-
return "0"
|
| 175 |
-
|
| 176 |
-
def analyze_calculation_complexity(operand_a: int, operand_b: int, operation: str) -> dict:
|
| 177 |
-
"""
|
| 178 |
-
Analyse la complexité d'un calcul pour enrichir les métadonnées dataset
|
| 179 |
-
|
| 180 |
-
Args:
|
| 181 |
-
operand_a: Premier opérande
|
| 182 |
-
operand_b: Deuxième opérande
|
| 183 |
-
operation: Type d'opération (×, +, -, ÷)
|
| 184 |
-
|
| 185 |
-
Returns:
|
| 186 |
-
Dictionnaire avec score de complexité et catégorie
|
| 187 |
-
"""
|
| 188 |
-
complexity_score = 0
|
| 189 |
-
|
| 190 |
-
if operation == "×":
|
| 191 |
-
complexity_score = max(operand_a, operand_b)
|
| 192 |
-
elif operation == "+":
|
| 193 |
-
complexity_score = (operand_a + operand_b) / 20
|
| 194 |
-
elif operation == "-":
|
| 195 |
-
complexity_score = max(operand_a, operand_b) / 10
|
| 196 |
-
elif operation == "÷":
|
| 197 |
-
complexity_score = operand_a / 10
|
| 198 |
-
|
| 199 |
-
# Catégorisation
|
| 200 |
-
if complexity_score < 5:
|
| 201 |
-
category = "easy"
|
| 202 |
-
elif complexity_score < 10:
|
| 203 |
-
category = "medium"
|
| 204 |
-
else:
|
| 205 |
-
category = "hard"
|
| 206 |
-
|
| 207 |
-
return {
|
| 208 |
-
"complexity_score": round(complexity_score, 2),
|
| 209 |
-
"difficulty_category": category,
|
| 210 |
-
"operation_type": operation
|
| 211 |
-
}
|
|
|
|
| 1 |
# ==========================================
|
| 2 |
+
# utils.py - Fonctions communes GPU simplifiées
|
| 3 |
# ==========================================
|
| 4 |
|
| 5 |
"""
|
| 6 |
+
Utilitaires partagés pour le traitement d'images OCR GPU
|
|
|
|
| 7 |
"""
|
| 8 |
|
| 9 |
from PIL import Image, ImageEnhance
|
| 10 |
import numpy as np
|
| 11 |
import gc
|
| 12 |
import os
|
|
|
|
| 13 |
|
| 14 |
def create_white_canvas(width: int = 300, height: int = 300) -> Image.Image:
|
| 15 |
"""Crée un canvas blanc pour le dessin de calculs"""
|
|
|
|
| 26 |
pass
|
| 27 |
|
| 28 |
def cleanup_memory() -> None:
|
| 29 |
+
"""Force le nettoyage mémoire GPU"""
|
| 30 |
gc.collect()
|
| 31 |
+
try:
|
| 32 |
+
import torch
|
| 33 |
+
if torch.cuda.is_available():
|
| 34 |
+
torch.cuda.empty_cache()
|
| 35 |
+
except:
|
| 36 |
+
pass
|
| 37 |
|
| 38 |
def optimize_image_for_ocr(image_dict: dict | np.ndarray | Image.Image | None, max_size: int = 300) -> Image.Image | None:
|
| 39 |
"""
|
| 40 |
+
Optimisation image pour OCR GPU
|
| 41 |
|
| 42 |
Args:
|
| 43 |
image_dict: Image d'entrée (format Gradio, numpy ou PIL)
|
|
|
|
| 125 |
size: Taille de la miniature (largeur, hauteur)
|
| 126 |
|
| 127 |
Returns:
|
| 128 |
+
HTML img tag avec image base64 ou icône par défaut
|
| 129 |
"""
|
| 130 |
try:
|
| 131 |
if optimized_image is None:
|
| 132 |
return "📝"
|
| 133 |
|
| 134 |
# Pour l'affichage dans l'interface, on garde le base64 temporairement
|
|
|
|
| 135 |
import base64
|
| 136 |
from io import BytesIO
|
| 137 |
|
|
|
|
| 174 |
# Si trop long, prendre les premiers chiffres
|
| 175 |
return cleaned_result[:max_length]
|
| 176 |
else:
|
| 177 |
+
return "0"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|