Create app.py

app.py

@@ -0,0 +1,346 @@
+import gradio as gr
+import numpy as np
+import cv2
+from PIL import Image
+from ultralytics import YOLO
+
+# ── Model ─────────────────────────────────────────────────────────────────────
+model = YOLO("best.pt")
+
+CLASS_NAMES  = {0: "Full",   1: "Broken"}
+CLASS_COLORS = {0: (34, 197, 94), 1: (239, 68, 68)}   # green, red
+
+
+# ── Inference ─────────────────────────────────────────────────────────────────
+def predict(image: Image.Image):
+    if image is None:
+        return None, "", ""
+
+    img_np  = np.array(image)
+    h, w    = img_np.shape[:2]
+    results = model(img_np, imgsz=1280, conf=0.25)[0]
+
+    annotated = img_np.copy()
+    overlay   = img_np.copy()
+
+    counts = {"Full": 0, "Broken": 0}
+
+    if results.masks is not None:
+        # Adaptive font scale based on image size
+        font_scale  = max(0.35, min(0.65, w / 2000))
+        font_thick  = 1
+        font        = cv2.FONT_HERSHEY_SIMPLEX
+
+        for mask_tensor, box in zip(results.masks.data, results.boxes):
+            cls_id   = int(box.cls[0])
+            cls_name = CLASS_NAMES.get(cls_id, "?")
+            color    = CLASS_COLORS.get(cls_id, (200, 200, 200))
+
+            counts[cls_name] += 1
+
+            # Resize mask to image size
+            mask_np = mask_tensor.cpu().numpy().astype(np.uint8)
+            mask_np = cv2.resize(mask_np, (w, h), interpolation=cv2.INTER_NEAREST)
+
+            # Fill overlay
+            overlay[mask_np == 1] = color
+
+            # Draw contour
+            cnts, _ = cv2.findContours(mask_np, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+            cv2.drawContours(annotated, cnts, -1, color, 2)
+
+            # Label placement — centroid of mask
+            ys, xs  = np.where(mask_np == 1)
+            if len(xs) == 0:
+                continue
+            cx = int(xs.mean())
+            cy = int(ys.mean())
+
+            label     = cls_name
+            (tw, th), baseline = cv2.getTextSize(label, font, font_scale, font_thick)
+
+            # Small pill background behind text
+            pad = 3
+            cv2.rectangle(annotated,
+                          (cx - tw // 2 - pad, cy - th - pad),
+                          (cx + tw // 2 + pad, cy + pad),
+                          (0, 0, 0), -1)
+            cv2.putText(annotated, label,
+                        (cx - tw // 2, cy),
+                        font, font_scale, color, font_thick, cv2.LINE_AA)
+
+        # Blend mask overlay with original
+        annotated = cv2.addWeighted(annotated, 0.72, overlay, 0.28, 0)
+
+        # Redraw contours on top of blend so they stay sharp
+        if results.masks is not None:
+            for mask_tensor, box in zip(results.masks.data, results.boxes):
+                cls_id  = int(box.cls[0])
+                color   = CLASS_COLORS.get(cls_id, (200, 200, 200))
+                mask_np = mask_tensor.cpu().numpy().astype(np.uint8)
+                mask_np = cv2.resize(mask_np, (w, h), interpolation=cv2.INTER_NEAREST)
+                cnts, _ = cv2.findContours(mask_np, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+                cv2.drawContours(annotated, cnts, -1, color, 2)
+
+    total   = counts["Full"] + counts["Broken"]
+    summary = f"**Total:** {total}   🟢 **Full:** {counts['Full']}   🔴 **Broken:** {counts['Broken']}"
+
+    # Table markdown
+    table = f"""| | Count |
+|---|---|
+| 🌾 Total Grains | **{total}** |
+| 🟢 Full Grains | **{counts['Full']}** |
+| 🔴 Broken Grains | **{counts['Broken']}** |"""
+
+    return Image.fromarray(annotated), summary, table
+
+
+# ── Custom CSS ────────────────────────────────────────────────────────────────
+css = """
+@import url('https://fonts.googleapis.com/css2?family=DM+Serif+Display&family=DM+Sans:wght@300;400;500&display=swap');
+
+* { box-sizing: border-box; }
+
+body, .gradio-container {
+    background: #0c0f0a !important;
+    font-family: 'DM Sans', sans-serif !important;
+}
+
+.gradio-container {
+    max-width: 1100px !important;
+    margin: 0 auto !important;
+    padding: 0 24px !important;
+}
+
+/* Header */
+#header {
+    text-align: center;
+    padding: 48px 0 32px;
+    border-bottom: 1px solid #1e2a1a;
+    margin-bottom: 36px;
+}
+
+#header h1 {
+    font-family: 'DM Serif Display', serif !important;
+    font-size: 2.6rem;
+    color: #e8f5e1;
+    letter-spacing: -0.5px;
+    margin: 0 0 10px;
+}
+
+#header p {
+    color: #6b8f5e;
+    font-size: 1rem;
+    font-weight: 300;
+    margin: 0;
+    letter-spacing: 0.3px;
+}
+
+/* Accent pill */
+#header span {
+    display: inline-block;
+    background: #1a2e14;
+    color: #7ec86a;
+    font-size: 0.72rem;
+    font-weight: 500;
+    letter-spacing: 1.5px;
+    text-transform: uppercase;
+    padding: 4px 14px;
+    border-radius: 20px;
+    margin-bottom: 18px;
+    border: 1px solid #2d4a24;
+}
+
+/* Panels */
+.panel-box {
+    background: #111710 !important;
+    border: 1px solid #1e2a1a !important;
+    border-radius: 12px !important;
+    padding: 0 !important;
+    overflow: hidden;
+}
+
+/* Input/output image areas */
+.image-wrap {
+    border-radius: 10px;
+    overflow: hidden;
+    background: #0c0f0a;
+}
+
+/* Upload button */
+.upload-btn, button[class*="upload"] {
+    background: #1a2e14 !important;
+    color: #7ec86a !important;
+    border: 1px dashed #2d4a24 !important;
+    border-radius: 10px !important;
+    font-family: 'DM Sans', sans-serif !important;
+}
+
+/* Submit button */
+#run-btn button, button#run-btn {
+    background: #3d7a2e !important;
+    color: #e8f5e1 !important;
+    font-family: 'DM Serif Display', serif !important;
+    font-size: 1.05rem !important;
+    letter-spacing: 0.3px !important;
+    border: none !important;
+    border-radius: 8px !important;
+    padding: 12px 0 !important;
+    transition: background 0.2s ease !important;
+    width: 100% !important;
+}
+
+#run-btn button:hover {
+    background: #4e9939 !important;
+}
+
+/* Summary text */
+#summary-box {
+    background: #111710 !important;
+    border: 1px solid #1e2a1a !important;
+    border-radius: 10px !important;
+    padding: 16px 20px !important;
+    color: #b5d4a8 !important;
+    font-size: 1rem !important;
+    font-family: 'DM Sans', sans-serif !important;
+    min-height: 52px !important;
+}
+
+/* Table */
+#table-box {
+    background: #111710 !important;
+    border: 1px solid #1e2a1a !important;
+    border-radius: 10px !important;
+    padding: 4px 0 !important;
+    font-family: 'DM Sans', sans-serif !important;
+}
+
+#table-box table {
+    width: 100% !important;
+    border-collapse: collapse !important;
+}
+
+#table-box th {
+    background: #182413 !important;
+    color: #6b8f5e !important;
+    font-size: 0.72rem !important;
+    font-weight: 500 !important;
+    letter-spacing: 1.2px !important;
+    text-transform: uppercase !important;
+    padding: 10px 18px !important;
+    border-bottom: 1px solid #1e2a1a !important;
+}
+
+#table-box td {
+    padding: 12px 18px !important;
+    color: #c8e0bf !important;
+    font-size: 0.95rem !important;
+    border-bottom: 1px solid #141d10 !important;
+}
+
+#table-box tr:last-child td {
+    border-bottom: none !important;
+}
+
+#table-box tr:hover td {
+    background: #141d10 !important;
+}
+
+/* Section labels */
+.section-label {
+    color: #4a6e3e;
+    font-size: 0.7rem;
+    font-weight: 500;
+    letter-spacing: 1.4px;
+    text-transform: uppercase;
+    margin-bottom: 8px;
+    padding: 0 2px;
+}
+
+/* Footer */
+#footer {
+    text-align: center;
+    color: #2d4224;
+    font-size: 0.78rem;
+    padding: 32px 0 24px;
+    border-top: 1px solid #141d10;
+    margin-top: 40px;
+    letter-spacing: 0.3px;
+}
+
+/* Gradio component overrides */
+.gr-block, .gr-box { background: transparent !important; border: none !important; }
+label.svelte-1b6s6vi, .gr-input-label { color: #4a6e3e !important; font-size: 0.7rem !important; letter-spacing: 1.2px !important; text-transform: uppercase !important; font-weight: 500 !important; }
+"""
+
+
+# ── Layout ────────────────────────────────────────────────────────────────────
+with gr.Blocks(css=css, title="GrainVision — Rice Grain Classifier") as demo:
+
+    gr.HTML("""
+    <div id="header">
+        <span>AI · Segmentation · Classification</span>
+        <h1>🌾 GrainVision</h1>
+        <p>Upload a rice image to detect and classify each grain as Full or Broken</p>
+    </div>
+    """)
+
+    with gr.Row(equal_height=False):
+
+        # ── Left: Input ───────────────────────────────────────────────────────
+        with gr.Column(scale=1):
+            gr.HTML('<div class="section-label">Input Image</div>')
+            input_img = gr.Image(
+                type="pil",
+                label="",
+                elem_classes=["image-wrap"],
+                height=420,
+            )
+            gr.HTML('<div style="height:12px"></div>')
+            run_btn = gr.Button("Analyse Grains", elem_id="run-btn")
+
+        # ── Right: Output ─────────────────────────────────────────────────────
+        with gr.Column(scale=1):
+            gr.HTML('<div class="section-label">Segmentation Result</div>')
+            output_img = gr.Image(
+                type="pil",
+                label="",
+                elem_classes=["image-wrap"],
+                height=420,
+            )
+
+    gr.HTML('<div style="height:20px"></div>')
+
+    # ── Results row ───────────────────────────────────────────────────────────
+    with gr.Row():
+        with gr.Column(scale=1):
+            gr.HTML('<div class="section-label">Detection Summary</div>')
+            summary_md = gr.Markdown(
+                value="Results will appear here after analysis.",
+                elem_id="summary-box",
+            )
+
+        with gr.Column(scale=1):
+            gr.HTML('<div class="section-label">Grain Count Table</div>')
+            table_md = gr.Markdown(
+                value="| | Count |\n|---|---|\n| 🌾 Total Grains | — |\n| 🟢 Full Grains | — |\n| 🔴 Broken Grains | — |",
+                elem_id="table-box",
+            )
+
+    # ── Events ────────────────────────────────────────────────────────────────
+    run_btn.click(
+        fn      = predict,
+        inputs  = [input_img],
+        outputs = [output_img, summary_md, table_md],
+    )
+    input_img.change(
+        fn      = predict,
+        inputs  = [input_img],
+        outputs = [output_img, summary_md, table_md],
+    )
+
+    gr.HTML('<div id="footer">GrainVision · Powered by YOLO11x-seg · For research & quality inspection use</div>')
+
+
+if __name__ == "__main__":
+    demo.launch()