inference.py

import tensorflow as tf
from tensorflow.keras.models import load_model
from huggingface_hub import hf_hub_download
import numpy as np
from PIL import Image

# ------------------------------
# Load the generator model from Hugging Face Hub
# ------------------------------
model_path = hf_hub_download(
    repo_id="saraayum/sar-colorization-generator",  # replace with your actual username/repo
    filename="sar_colorization_generator_final.keras"
)
# generator = load_model(model_path, compile=False)
generator = load_model(model_path, compile=False, safe_mode=False)
# ------------------------------
# Preprocess the SAR image
# ------------------------------
def preprocess_image(image_path):
    img = tf.io.read_file(image_path)
    img = tf.image.decode_png(img, channels=1)       # SAR = grayscale
    img = tf.image.resize(img, [256, 256])
    img = tf.cast(img, tf.float32) / 127.5 - 1.0     # Normalize to [-1, 1]
    img = tf.expand_dims(img, 0)                     # Add batch dimension
    return img

# ------------------------------
# Postprocess the generator output
# ------------------------------
def postprocess_output(output_tensor):
    output_tensor = (output_tensor[0] + 1.0) * 127.5     # Convert back to [0,255]
    output_tensor = tf.clip_by_value(output_tensor, 0, 255)
    output_image = tf.cast(output_tensor, tf.uint8)
    return Image.fromarray(output_image.numpy())

# ------------------------------
# Predict function for inference
# ------------------------------
def predict(image_path, save_path="output.png"):
    sar_input = preprocess_image(image_path)
    gen_output = generator(sar_input, training=False)
    output_image = postprocess_output(gen_output)
    output_image.save(save_path)
    print(f"Colorized image saved as: {save_path}")
    return output_image

# ------------------------------
# Example usage (you can comment this out on HF)
# ------------------------------
if __name__ == "__main__":
    predict("sample_sar.png", "predicted_colorized.png")