app.py

import gradio as gr
from ultralytics import YOLO
import cv2
import numpy as np
import supervision as sv
import os
from pathlib import Path 
from supervision.draw.color import Color


# Create title, description and article strings
title = "Excavator Detector"
description = "An Yolov8(m) object detection computer vision model to detect images of excavators."

excavator_path = Path('.')
example_path = excavator_path / "examples"

# Get example filepaths in a list of lists
example_list = [["examples/" + example] for example in os.listdir(example_path)]

def predict(img, conf_thres=0.35, iou_thres=0.5): # default : conf(0.25), iou(0.7)
    model = YOLO('./weights/excavator_only_best_v8m.pt')
    classNames = ['excavator']
    
    results = model.predict(source=img, conf=conf_thres, iou=iou_thres)[0]
    detections = sv.Detections.from_yolov8(results)

    box_annotator = sv.BoxAnnotator()
    box_annotator.text_scale = 1.3
    box_annotator.text_color = Color.white()
    box_annotator.text_thickness = 2
    
    labels = [
        f"{classNames[class_id]} {confidence:.2f}"
        for class_id, confidence
        in zip(detections.class_id, detections.confidence)
    ]
    image = cv2.cvtColor(cv2.imread(img),cv2.COLOR_BGR2RGB)
    annotated_image = box_annotator.annotate(image, detections=detections, labels=labels)
    
    return annotated_image, results.speed



# Create the Gradio demo
demo = gr.Interface(fn=predict, # mapping function from input to output
                    inputs=gr.Image(type="filepath"), # what are the inputs?
                    outputs=[gr.Image(type='numpy'),
                             gr.Text(label="Prediction time (ms)")],
                    examples=example_list, 
                    title=title,
                    description=description)

# Launch the demo!
demo.launch(debug=False # print errors locally?
            ) # share=True, generate a publically shareable URL?