Update app.py

app.py

@@ -5,15 +5,17 @@ import gradio as gr
 import pandas as pd
 from ultralytics import YOLO
 from pathlib import Path
+from datetime import datetime
 
 # === Konfigurasi ===
-model_path = "best.pt"
-gt_label_dir = "data/labels"
+model_path = "best.pt"  # pastikan file ini ada di root repo Hugging Face
+gt_label_dir = "data/labels"  # sesuaikan dengan struktur foldermu
 class_names = ['coral or rock', 'pipeline', 'ripple marks', 'shipwreck']
 
+# Load model
 model = YOLO(model_path)
 
-# === Fungsi untuk menggambar mask ===
+# === Fungsi untuk menggambar segmentasi ===
 def draw_mask(img, segments, class_ids, color=(255, 0, 0), alpha=0.5, label_type="Pred", confs=None):
     overlay = img.copy()
     for i, seg in enumerate(segments):
@@ -22,7 +24,8 @@ def draw_mask(img, segments, class_ids, color=(255, 0, 0), alpha=0.5, label_type
         polygon[:, 1] *= img.shape[0]
         polygon = polygon.astype(np.int32)
 
-        label = f"{label_type}: {class_names[class_ids[i]]}"
+        cls_id = int(class_ids[i])  # FIX: pastikan integer
+        label = f"{label_type}: {class_names[cls_id]}"
         if confs and i < len(confs):
             label += f" ({confs[i]:.2f})"
 
@@ -32,14 +35,17 @@ def draw_mask(img, segments, class_ids, color=(255, 0, 0), alpha=0.5, label_type
 
     return cv2.addWeighted(overlay, alpha, img, 1 - alpha, 0)
 
-# === Proses gambar dan simpan report ===
+# === Proses prediksi + simpan CSV ===
 def process_image_and_report(image):
-    temp_path = "temp.jpg"
+    timestamp = datetime.now().strftime("%Y%m%d%H%M%S")
+    temp_path = f"/tmp/temp_{timestamp}.jpg"
+    csv_path = f"/tmp/detection_report_{timestamp}.csv"
+
     image.save(temp_path)
     img = cv2.imread(temp_path)
     img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
 
-    # === Ground Truth ===
+    # === Load ground truth ===
     label_path = os.path.join(gt_label_dir, Path(temp_path).with_suffix(".txt").name)
     img_gt = img.copy()
     if os.path.exists(label_path):
@@ -56,7 +62,7 @@ def process_image_and_report(image):
                 gt_ids.append(cls_id)
             img_gt = draw_mask(img_gt, gt_segments, gt_ids, color=(0, 255, 0), label_type="GT")
 
-    # === Prediction YOLOv8 ===
+    # === Deteksi dengan YOLOv8 ===
     results = model(temp_path)[0]
     segs = [seg.xy for seg in results.masks] if results.masks else []
     cls_ids = results.boxes.cls.tolist() if results.boxes else []
@@ -65,10 +71,10 @@ def process_image_and_report(image):
 
     img_pred = draw_mask(img.copy(), segs, cls_ids, color=(255, 0, 0), label_type="Pred", confs=confs)
 
-    # === Gabung visualisasi
+    # === Gabung GT dan Prediksi
     combined = np.concatenate((img_gt, img_pred), axis=1)
 
-    # === Buat laporan CSV
+    # === Simpan laporan ke CSV
     rows = []
     for i in range(len(cls_ids)):
         rows.append({
@@ -82,15 +88,14 @@ def process_image_and_report(image):
         })
 
     df = pd.DataFrame(rows)
-    csv_path = "detection_report.csv"
     df.to_csv(csv_path, index=False)
 
     return combined, csv_path
 
-# === Gradio UI ===
+# === UI Gradio ===
 with gr.Blocks() as demo:
     gr.Markdown("## 🧠 YOLOv8 Segmentasi Viewer + Report Generator")
-    gr.Markdown("Upload gambar, lihat segmentasi, dan unduh hasil deteksi sebagai CSV")
+    gr.Markdown("Upload gambar, lihat segmentasi prediksi vs GT, dan unduh laporan CSV deteksi.")
 
     with gr.Row():
         with gr.Column():