import os
import gradio as gr
from transformers import pipeline, DetrForObjectDetection, DetrConfig, DetrImageProcessor

# Initialize the configuration for DetrForObjectDetection
config = DetrConfig.from_pretrained("facebook/detr-resnet-50")

# Create the model for object detection using the specified configuration
model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50", config=config)

# Initialize the image processor for DETR
image_processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")

# Initialize the object detection pipeline with the model and image processor
od_pipe = pipeline(task='object-detection', model=model, image_processor=image_processor)

def get_pipeline_prediction(pil_image):
    # Run the object detection pipeline on the input image
    pipeline_output = od_pipe(pil_image)
    
    # You might need to implement or adjust the rendering function based on the `pipeline_output`
    # The `render_results_in_image` function is assumed here to draw bounding boxes and labels on the input image,
    # but you'll need to define it according to your specific needs.
    # For now, the output is directly returned since the question doesn't define `render_results_in_image`.
    return pipeline_output

demo = gr.Interface(
  fn=get_pipeline_prediction,
  inputs=gr.Image(label="Input image", 
                  type="pil"),
  outputs=gr.JSON(label="Detected objects")  # Adjusted to show JSON output if rendering function is not defined
)

demo.launch()