import os
import numpy as np
import tensorflow as tf
import time
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from tensorflow.keras.models import load_model
from PIL import Image
# Ladataan koulutettu malli
model_path = 'MobileNetV2_military_vehicles_model.tf'
model = load_model(model_path)
THUMBNAILS_DIR = 'thumbnails'
if not os.path.exists(THUMBNAILS_DIR):
os.makedirs(THUMBNAILS_DIR)
def create_thumbnail(image_path, thumbnail_size=(128, 128)):
img = Image.open(image_path)
img.thumbnail(thumbnail_size)
return img.convert("RGB") # Muutetaan kuva RGB-muotoon
def generate_html(predictions, analyzed_count, elapsed_time):
# Järjestä ennustukset confidencen mukaan suurimmasta pienimpään
sorted_predictions = sorted(predictions, key=lambda x: x[2], reverse=True)
html_content = f"""
Scan Results
Analysis results
Analyzed {analyzed_count} images in {elapsed_time:.2f} seconds.
| Image |
Filename |
Prediction |
"""
for prediction in sorted_predictions:
image_path, predicted_class, confidence = prediction
thumbnail = create_thumbnail(image_path)
filename, file_extension = os.path.splitext(os.path.basename(image_path))
thumbnail_filename = f"{filename}_thumb{file_extension}"
thumbnail_path = os.path.join(THUMBNAILS_DIR, thumbnail_filename)
thumbnail.save(thumbnail_path, "JPEG")
html_content += f" | {filename+file_extension} | {predicted_class} with confidence {confidence:.2f} |
"
html_content += "
"
with open("results-MobileNetV2.html", "w") as f:
f.write(html_content)
def preprocess_image(image_path):
img = load_img(image_path, target_size=(224, 224))
img_array = img_to_array(img)
img_array = np.expand_dims(img_array, axis=0)
return img_array / 255.0
def scan_directory(dir_to_scan):
sotilaskalusto_classes = [
'Anti-aircraft', 'Armored combat support vehicles', 'Armored personnel carriers',
'Infantry fighting vehicles', 'Light armored vehicles', 'Mine-protected vehicles',
'Prime movers and trucks', 'Self-propelled artillery', 'light utility vehicles', 'tanks'
]
predictions_list = []
analyzed_count = 0
start_time = time.time() # Aloita ajanotto
for root, dirs, files in os.walk(dir_to_scan):
for file in files:
if file.lower().endswith(('.png', '.jpg', '.jpeg')):
analyzed_count += 1 # Lisää analysoitu kuva laskuriin
image_path = os.path.join(root, file)
image = preprocess_image(image_path)
predictions = model.predict(image)
predicted_class = sotilaskalusto_classes[np.argmax(predictions)]
if np.max(predictions) > 0.01:
predictions_list.append((image_path, predicted_class, np.max(predictions)))
elapsed_time = time.time() - start_time # Laske kulunut aika
generate_html(predictions_list, analyzed_count, elapsed_time)
# dir_to_scan = './mil'
dir_to_scan = './social-media'
scan_directory(dir_to_scan)