app.py

from typing import List
import cv2
import torch
import numpy as np
import gradio as gr
import config as modelConfig

from pytorch_grad_cam.utils.image import show_cam_on_image

from yolov3 import YOLOv3
import utils
from utils import cells_to_bboxes, non_max_suppression, draw_bounding_boxes, YoloGradCAM


model = YOLOv3(num_classes=len(modelConfig.PASCAL_CLASSES))
optimizer = torch.optim.Adam(model.parameters(), lr=0.00072/100, weight_decay=1e-4)

utils.load_checkpoint("checkpoint.pth.tar", model, optimizer, 0.00072/100)

scaled_anchors = (
    torch.tensor(modelConfig.ANCHORS)
    * torch.tensor(modelConfig.S).unsqueeze(1).unsqueeze(1).repeat(1, 3, 2)
).to(modelConfig.DEVICE)

yolo_grad_cam = YoloGradCAM(model=model, target_layers=[model.layers[-2]], use_cuda=False)

@torch.inference_mode()
def detect_objects(image: np.ndarray, iou_thresh: float = 0.5, thresh: float = 0.4, enable_grad_cam: bool = False, transparency: float = 0.5) -> List[np.ndarray]:
    transformed_image = modelConfig.transforms(image=image)["image"].unsqueeze(0)
    #transformed_image = transformed_image.cuda()
    output = model(transformed_image)

    bboxes = [[] for _ in range(1)]
    for i in range(3):
        batch_size, A, S, _, _ = output[i].shape
        anchor = scaled_anchors[i]
        boxes_scale_i = cells_to_bboxes(
            output[i], anchor, S=S, is_preds=True
        )
        for idx, (box) in enumerate(boxes_scale_i):
            bboxes[idx] += box

    nms_boxes = non_max_suppression(
        bboxes[0], iou_threshold=iou_thresh, threshold=thresh, box_format="midpoint",
    )
    plot_img_with_bboxes = draw_bounding_boxes(image.copy(), nms_boxes, class_labels=modelConfig.PASCAL_CLASSES)
    if not enable_grad_cam:
        return [plot_img_with_bboxes]

    grayscale_cam = yolo_grad_cam(transformed_image, scaled_anchors)[0, :, :]
    img = cv2.resize(image, (416, 416))
    img = np.float32(img) / 255
    grad_cam_image = show_cam_on_image(img, grayscale_cam, use_rgb=True, image_weight=transparency)
    return [plot_img_with_bboxes, grad_cam_image]

def inference(
    image: np.ndarray,
    iou_thresh: float, thresh: float,
    enable_grad_cam: str,
    transparency: float,
):
    results = detect_objects(image, iou_thresh, thresh, enable_grad_cam, transparency)
    return results


title = "Object detection application using YoloV3 Model"
description = f"Object detection application using pre-trained YoloV3 model for Pascal VOC dataset. This app has GradCAM option also. \n The 20 classes in Pascal voc dataset are : {', '.join(modelConfig.PASCAL_CLASSES)}"
examples = [
    ["images/000811.jpg", 0.6, 0.6, True, 0.6],
    ["images/000830.jpg", 0.5, 0.5, True, 0.6],
    ["images/000842.jpg", 0.6, 0.6, True, 0.6],    
    ["images/001114.jpg", 0.4, 0.5, True, 0.6],
    ["images/001133.jpg", 0.7, 0.7, True, 0.6],
    ["images/001155.jpg", 0.7, 0.69, True, 0.6],
    ["images/000008.jpg", 0.66, 0.69, True, 0.6],
    ["images/000031.jpg", 0.6, 0.6, True, 0.6],
    ["images/000175.jpg", 0.6, 0.6, True, 0.6],
]

demo = gr.Interface(
    inference,
    inputs=[
        gr.Image(label="Input Image"),
        gr.Slider(0, 1, value=0.5, label="IOU Threshold"),
        gr.Slider(0, 1, value=0.4, label="Threshold"),
        gr.Checkbox(label="Show Grad Cam"),
        gr.Slider(0, 1, value=0.5, label="Opacity of GradCAM"),
    ],
    outputs=[
        gr.Gallery(rows=2, columns=1),
    ],
    title=title,
    description=description,
    examples=examples,
)

demo.launch(debug=True)