from transformers import AutoTokenizer, AutoImageProcessor, VisionEncoderDecoderModel
import requests, time
from PIL import Image

model_path = "cnmoro/nano-image-captioning"

# load the image captioning model and corresponding tokenizer and image processor
model = VisionEncoderDecoderModel.from_pretrained(model_path)
tokenizer = AutoTokenizer.from_pretrained(model_path)
image_processor = AutoImageProcessor.from_pretrained(model_path)

# preprocess an image
url = "https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/New_york_times_square-terabass.jpg/800px-New_york_times_square-terabass.jpg"
image = Image.open(requests.get(url, stream=True).raw)
pixel_values = image_processor(image, return_tensors="pt").pixel_values

start = time.time()

# generate caption - suggested settings
generated_ids = model.generate(
    pixel_values,
    temperature=0.7,
    top_p=0.8,
    top_k=50,
    num_beams=3 # you can use 1 for even faster inference with a small drop in quality
)
generated_text = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

end = time.time()

print(generated_text)
# a group of people are in the middle of a city.

print(f"Time taken: {end - start} seconds")
# Time taken: 0.07550048828125 seconds
# on CPU !