app.py

try:
    import detectron2
except:
    import os 
    os.system('pip install git+https://github.com/facebookresearch/detectron2.git')

import cv2

from matplotlib.pyplot import axis
import gradio as gr
import requests
import numpy as np
from torch import nn
import requests

import torch

from detectron2 import model_zoo
from detectron2.engine import DefaultPredictor
from detectron2.config import get_cfg
from detectron2.utils.visualizer import Visualizer
from detectron2.data import MetadataCatalog


model_path = "https://huggingface.co/dbmdz/detectron2-model/resolve/main/model_final.pth"

cfg = get_cfg()
cfg.merge_from_file("./configs/detectron2/faster_rcnn_R_50_FPN_3x.yaml")
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 2
cfg.MODEL.WEIGHTS = model_path

my_metadata = MetadataCatalog.get("dbmdz_coco_all")
my_metadata.thing_classes = ["Illumination", "Illustration"]

if not torch.cuda.is_available():
    cfg.MODEL.DEVICE = "cpu"


def inference(image_url, image, min_score):
    if image_url:
        r = requests.get(image_url)
        if r:
            im = np.frombuffer(r.content, dtype="uint8")
            im = cv2.imdecode(im, cv2.IMREAD_COLOR)
    else:
        # Model expect BGR!
        im = image[:,:,::-1]

    cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = min_score
    predictor = DefaultPredictor(cfg)

    outputs = predictor(im)

    v = Visualizer(im, my_metadata, scale=1.2)
    out = v.draw_instance_predictions(outputs["instances"].to("cpu"))

    return out.get_image()


title = "DBMDZ Detectron2 Model Demo"
description = "This demo introduces an interactive playground for our trained Detectron2 model. <br>The model was trained on manually annotated segments from digitized books to detect Illustration or Illumination segments on a given page."
article = '<p>Detectron model is available from our repository <a href="">here</a> on the Hugging Face Model Hub.</p>'

gr.Interface(
    inference,
    [gr.inputs.Textbox(label="Image URL", placeholder="https://api.digitale-sammlungen.de/iiif/image/v2/bsb10483966_00008/full/500,/0/default.jpg"),
     gr.inputs.Image(type="numpy", label="Input Image"),
     gr.Slider(minimum=0.0, maximum=1.0, value=0.5, label="Minimum score:"),
    ], 
    gr.outputs.Image(type="pil", label="Output"),
    title=title,
    description=description,
    article=article,
    examples=[]).launch()