| from transformers import pipeline | |
| od_pipe = pipeline("object-detection", "facebook/detr-resnet-50") | |
| import os | |
| import gradio as gr | |
| from helper import render_results_in_image | |
| def get_pipeline_prediction(pil_image): | |
| pipeline_output = od_pipe(pil_image) | |
| processed_image = render_results_in_image(pil_image, | |
| pipeline_output) | |
| return processed_image | |
| object_detection_app = gr.Interface( | |
| fn=get_pipeline_prediction, | |
| inputs=gr.Image(label="Input image", | |
| type="pil"), | |
| outputs=gr.Image(label="Output image with predicted instances", | |
| type="pil"), | |
| title = "Object Detection", | |
| description = "Facebook detr-resnet-50") | |
| with gr.Blocks() as demo: | |
| gr.TabbedInterface( | |
| [object_detection_app], | |
| ["Object Detection"]) | |
| demo.launch() | |