Create app.py

app.py

@@ -0,0 +1,109 @@
+import gradio as gr
+from PIL import Image
+
+# Use a pipeline as a high-level helper
+from transformers import pipeline
+
+pipe = pipeline("object-detection", model="facebook/detr-resnet-101")
+
+from PIL import Image, ImageDraw, ImageFont
+
+def merge_bboxes(bboxes, threshold=50):
+    """
+    Merges bounding boxes that are close to each other into one.
+
+    :param bboxes: List of bounding boxes (xmin, ymin, xmax, ymax)
+    :param threshold: Maximum distance between boxes to be considered as part of the same object
+    :return: List of merged bounding boxes
+    """
+    merged_bboxes = []
+    for bbox in bboxes:
+        added = False
+        for i, merged_bbox in enumerate(merged_bboxes):
+            if (abs(bbox[0] - merged_bbox[0]) < threshold and abs(bbox[1] - merged_bbox[1]) < threshold) or \
+               (abs(bbox[2] - merged_bbox[2]) < threshold and abs(bbox[3] - merged_bbox[3]) < threshold):
+                # Merge the boxes by extending the boundaries
+                new_bbox = (
+                    min(bbox[0], merged_bbox[0]),
+                    min(bbox[1], merged_bbox[1]),
+                    max(bbox[2], merged_bbox[2]),
+                    max(bbox[3], merged_bbox[3])
+                )
+                merged_bboxes[i] = new_bbox
+                added = True
+                break
+        if not added:
+            merged_bboxes.append(bbox)
+    return merged_bboxes
+
+def draw_bounding_boxes(image, detections):
+    """
+    Draws bounding boxes on the given PIL image based on the object detection results.
+    Merges close bounding boxes to avoid duplicate detections.
+
+    :param image: PIL Image object
+    :param detections: List of dictionaries containing object detection results
+    :return: PIL Image with bounding boxes
+    """
+    draw = ImageDraw.Draw(image)
+
+    try:
+        font = ImageFont.truetype("arial.ttf", 40)  # Load font (adjust size as needed)
+    except:
+        font = ImageFont.load_default()  # Use default font if Arial is not available
+
+    bboxes = []
+    labels = []
+
+    # Collect all bounding boxes and labels
+    for obj in detections:
+        box = obj["box"]
+        label = f"{obj['label']} ({obj['score']:.2f})"
+        xmin, ymin, xmax, ymax = box["xmin"], box["ymin"], box["xmax"], box["ymax"]
+
+        bboxes.append((xmin, ymin, xmax, ymax))
+        labels.append(label)
+
+    # Merge close bounding boxes
+    merged_bboxes = merge_bboxes(bboxes)
+
+    # Draw bounding boxes and labels for merged boxes
+    for idx, bbox in enumerate(merged_bboxes):
+        xmin, ymin, xmax, ymax = bbox
+        label = labels[idx]
+
+        # Draw bounding box
+        draw.rectangle([xmin, ymin, xmax, ymax], outline="red", width=5)
+
+        # Adjust label placement to avoid overlap
+        label_bbox = draw.textbbox((xmin, ymin), label, font=font)
+        label_width, label_height = label_bbox[2] - label_bbox[0], label_bbox[3] - label_bbox[1]
+        label_x = xmin
+        label_y = ymin - label_height if ymin - label_height > 0 else ymin + 5  # Adjust above or below box
+
+        # Draw label background to make text more visible
+        draw.rectangle([label_x, label_y, label_x + label_width, label_y + label_height], fill="red")
+        draw.text((label_x, label_y), label, fill="white", font=font)
+
+    return image
+
+# Example usage:
+# image = Image.open("your_image.jpg")
+# detections = [{'score': 0.92, 'label': 'dog', 'box': {'xmin': 2929, 'ymin': 1297, 'xmax': 5067, 'ymax': 3693}}, ...]
+# image_with_boxes = draw_bounding_boxes(image, detections)
+# image_with_boxes.show()
+
+def detect_object(image):
+    raw_imagee = image
+    output=pipe(raw_imagee)
+    processed_image = draw_bounding_boxes(raw_imagee, output)
+    return processed_image
+
+demo = gr.Interface(fn = detect_object,
+                   inputs=[gr.Image(label='Please upload the file to detect the data', type='pil')],
+                   outputs=[gr.Image(label='Detected object', type='pil')],
+                   title= 'Object Detector',
+                   description = 'This application will be used to detect image'
+                   )
+
+demo.launch(share=True)