Update app.py

app.py

@@ -1,54 +1,54 @@
-from ultralytics import YOLO
-import gradio as gr
-from PIL import Image
-import torch
-
-# Load model
-model = YOLO("C:\\Users\\warho\\Desktop\\bonsAI\\best.pt")  # make sure best.pt is in the same folder
-
-# Prediction function
-def predict(inp):
-    if inp is None:
-        return None, {}
-    results = model.predict(source=inp, conf=0.5, iou=0.5, imgsz=640)
-    r = results[0]
-    output_img = r.plot()[:, :, ::-1]  # convert BGR to RGB for Gradio
-
-    # Build confidence dictionary (like the example video)
-    conf_dict = {}
-    for box in r.boxes:
-        cls_id = int(box.cls.item())
-        cls_name = model.names[cls_id]
-        conf = round(float(box.conf.item()) * 100, 2)
-        conf_dict[cls_name] = conf
-
-    return Image.fromarray(output_img), conf_dict
-
-
-# Gradio interface (similar style to video)
-demo = gr.Interface(
-    fn=predict,
-    inputs=gr.Image(type="pil", label="Upload Pill Image"),
-    outputs=[
-        gr.Image(type="pil", label="Detected Pills"),
-        gr.Label(num_top_classes=3, label="Predictions")
-    ],
-    title="bonsAI Pill Detection",
-    description=(
-        "Upload an image of a pill. The YOLOv12 model detects and classifies 20 pill types "
-        "commonly found in the Philippines. This study aims to automate pill recognition for "
-        "pharmaceutical verification and healthcare support using computer vision and deep learning."
-    ),
-    article=(
-        "### Study Summary\n"
-        "The bonsAI project demonstrates the application of YOLOv12 in real-time pill classification "
-        "and segmentation. By training on the Pharmaceutical Drugs and Vitamins Dataset (Version 2), "
-        "the system accurately identifies tablets and capsules across 20 classes using bounding boxes "
-        "and mask segmentation. The model achieved high mAP and F1-scores, confirming its potential "
-        "for aiding pharmacists and healthcare providers in ensuring drug authenticity and preventing "
-        "dispensing errors."
-    )
-)
-
-
-demo.launch()
+from ultralytics import YOLO
+import gradio as gr
+from PIL import Image
+import torch
+
+# Load model
+model = YOLO("best.pt")  # make sure best.pt is in the same folder
+
+# Prediction function
+def predict(inp):
+    if inp is None:
+        return None, {}
+    results = model.predict(source=inp, conf=0.5, iou=0.5, imgsz=640)
+    r = results[0]
+    output_img = r.plot()[:, :, ::-1]  # convert BGR to RGB for Gradio
+
+    # Build confidence dictionary (like the example video)
+    conf_dict = {}
+    for box in r.boxes:
+        cls_id = int(box.cls.item())
+        cls_name = model.names[cls_id]
+        conf = round(float(box.conf.item()) * 100, 2)
+        conf_dict[cls_name] = conf
+
+    return Image.fromarray(output_img), conf_dict
+
+
+# Gradio interface (similar style to video)
+demo = gr.Interface(
+    fn=predict,
+    inputs=gr.Image(type="pil", label="Upload Pill Image"),
+    outputs=[
+        gr.Image(type="pil", label="Detected Pills"),
+        gr.Label(num_top_classes=3, label="Predictions")
+    ],
+    title="bonsAI Pill Detection",
+    description=(
+        "Upload an image of a pill. The YOLOv12 model detects and classifies 20 pill types "
+        "commonly found in the Philippines. This study aims to automate pill recognition for "
+        "pharmaceutical verification and healthcare support using computer vision and deep learning."
+    ),
+    article=(
+        "### Study Summary\n"
+        "The bonsAI project demonstrates the application of YOLOv12 in real-time pill classification "
+        "and segmentation. By training on the Pharmaceutical Drugs and Vitamins Dataset (Version 2), "
+        "the system accurately identifies tablets and capsules across 20 classes using bounding boxes "
+        "and mask segmentation. The model achieved high mAP and F1-scores, confirming its potential "
+        "for aiding pharmacists and healthcare providers in ensuring drug authenticity and preventing "
+        "dispensing errors."
+    )
+)
+
+
+demo.launch()