import gradio as gr
from transformers import AutoImageProcessor, TableTransformerForObjectDetection
import torch
from PIL import Image

def inference(img):
    pretrained_model_path = "table-transformer-detection"

    image = img.convert("RGB")

    image_processor = AutoImageProcessor.from_pretrained(pretrained_model_path)
    model = TableTransformerForObjectDetection.from_pretrained(pretrained_model_path)

    inputs = image_processor(images=image, return_tensors="pt")
    outputs = model(**inputs)

    # convert outputs (bounding boxes and class logits) to COCO API
    target_sizes = torch.tensor([image.size[::-1]])
    results = image_processor.post_process_object_detection(outputs, threshold=0.9, target_sizes=target_sizes)[
        0
    ]

    for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
        box = [round(i, 2) for i in box.tolist()]
        return (
            f"Detected {model.config.id2label[label.item()]} with confidence "
            f"{round(score.item(), 3)} at location {box}"
        )


title = "object detection:table-transformer-detection"
description = "Gradio Demo for table-transformer-detection. To use it, simply upload your image, or click one of the examples to load them."
article = "<p style='text-align: center'><a href='https://github.com/bryandlee/animegan2-pytorch' target='_blank'>Github Repo Pytorch</a></p> <center><img src='https://visitor-badge.glitch.me/badge?page_id=akhaliq_animegan' alt='visitor badge'></center></p>"
examples=[['example_pdf.png']]


demo = gr.Interface(
    fn=inference,
    inputs=[gr.inputs.Image(type="pil")],
    outputs=gr.outputs.Textbox(),
    title=title,
    description=description,
    article=article,
    examples=examples)

demo.launch()