embed.py

import argparse
import csv
import os

from PIL import Image

from utils import load_model


def main(args):
    root = args.image_path
    files = list(os.listdir(root))
    for model_name in ["koclip", "koclip/koclip-large"]:
        counter = 0
        images = []
        image_ids = []
        model, processor = load_model(f"koclip/{model_name}")
        while counter < len(files):
            if counter != 0 and counter % args.batch_size == 0:
                inputs = processor(text=[""], images=images, return_tensors="jax", padding=True)
                features = model(**inputs).image_embeds
                with open(os.path.join(args.out_path, f"{model_name}.tsv", "w+")) as f:
                    writer = csv.writer(f, delimiter="\t")
                    for image_id, feature in zip(image_ids, features):
                        writer.writerow([image_id, ",".join(feature)])
                images = []
                image_ids = []
            else:
                file_ = files[counter]
                image = Image.open(os.path.join(root, file_))
                images.append(image)
                image_ids.append(file_)
                counter += 1
        


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--batch_size", default=16)
    parser.add_argument("--image_path", default="images")
    parser.add_argument("--out_path", default="features")
    args = parser.parse_args()
    main(args)