# Image Captioning from:
# https://learn.deeplearning.ai/courses/open-source-models-hugging-face/lesson/12/image-captioning
#

from transformers import BlipForConditionalGeneration
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")

from transformers import AutoProcessor
processor = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")

from PIL import Image
import gradio as gr

def captioning(input):
    image_tensors = processor(input, return_tensors="pt")
    image_text_tensors = model.generate(**image_tensors)
    output = processor.decode(image_text_tensors[0], skip_special_tokens=True)
    return output

gr.close_all()

app = gr.Interface(fn=captioning,
                   inputs=[gr.Image(label="Laita tähä joku kuva", type="pil")],
                   outputs=[gr.Textbox(label="Mitä näkyy?")],
                   title="Harzan kuvan selitys aplikaatio",
                   description="Harzan ihme aplikaatio kertomaan mitä kuvassa on",
                   allow_flagging="never")
app.launch()
gr.close_all()