File size: 852 Bytes
cc7d279 830c435 cc7d279 9f43ed3 1c8dc00 cc7d279 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
import tensorflow as tf
import numpy as np
from PIL import Image
import os
MODEL_PATH = os.path.join(
os.path.dirname(__file__),
"saved_model",
"Inception_V3_Animals_Classification.h5"
)
model = tf.keras.models.load_model(MODEL_PATH)
CLASS_NAMES = ["Cat", "Dog", "Snake"]
def preprocess_image(img: Image.Image, target_size=(256, 256)):
img = img.convert("RGB")
img = img.resize(target_size)
img = np.array(img).astype("float32") / 255.0
img = np.expand_dims(img, axis=0)
return img
def predict(img: Image.Image):
input_tensor = preprocess_image(img)
preds = model.predict(input_tensor)[0]
class_idx = int(np.argmax(preds))
confidence = float(np.max(preds))
prob_dict = {CLASS_NAMES[i]: float(preds[i]) for i in range(len(CLASS_NAMES))}
return CLASS_NAMES[class_idx], confidence, prob_dict |