app.py

import gradio as gr
from PIL import Image
from pathlib import Path
import os

from torchvision.io.image import read_image
from torchvision.models.detection import fasterrcnn_resnet50_fpn_v2, FasterRCNN_ResNet50_FPN_V2_Weights
from torchvision.utils import draw_bounding_boxes
from torchvision.transforms.functional import to_pil_image
import torch

weights = FasterRCNN_ResNet50_FPN_V2_Weights.DEFAULT
model = fasterrcnn_resnet50_fpn_v2(weights=weights, box_score_thresh=0.9)
#path = Path("object_detection_model.pth")
#model = torch.load(path)

from PIL import Image,ImageDraw
from torchvision.io.image import read_image
from torchvision.models.detection import fasterrcnn_resnet50_fpn_v2, FasterRCNN_ResNet50_FPN_V2_Weights
from torchvision.utils import draw_bounding_boxes
from torchvision import transforms
from torchvision.transforms.functional import to_pil_image,to_tensor
from pathlib import Path
import torch

# Step 1: Initiali1ze model with the best available weights
weights = FasterRCNN_ResNet50_FPN_V2_Weights.DEFAULT
model = fasterrcnn_resnet50_fpn_v2(weights=weights, box_score_thresh=0.9)
preprocess = weights.transforms()

def predict1(inp1):
  model.eval()

  batch = [preprocess(inp1)]

  # Step 4: Use the model and visualize the prediction
  prediction = model(batch)[0]
  labels = [weights.meta["categories"][i] for i in prediction["labels"]]
  
  draw = ImageDraw.Draw(inp1)
  z = prediction["boxes"]

  # Define the bounding box coordinates and size (adjust these values as needed)
  outline_color = "red"
  line_width = 2

  # Draw the bounding box as a rectangle
  x,y,l,k = z[0][0].item(), z[0][1].item(), z[0][2].item(), z[0][3].item()
  draw.rectangle([x, y,l,k], outline=outline_color, width=line_width)

  return labels[0],inp1


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

title = "Object Detection Model😀😀"
description = "A Object Detection model based on pretrained resnet neural network . It can detect car,airplane,bus,boat,train, bird, horse, suitcase, handbag etc."
article = "Created by [Rounak Gera](https://github.com/rounak890)"

demo = gr.Interface(fn=predict1, inputs=[gr.components.Image(type="pil", label="Image 1")],
                    outputs= [gr.components.Label(num_top_classes = 4, label = "Predictions"),
                              gr.components.Image(type = "pil",label = "predicted")],
                    title = title,
                    description = description,
                    article = article,
                    examples = example_list)
demo.launch()