Update app.py

app.py

@@ -22,12 +22,46 @@ MODEL_CANDIDATES = []
 if HF_MODEL_REPO and HF_MODEL_FILE:
     MODEL_CANDIDATES.append((HF_MODEL_REPO, HF_MODEL_FILE))
 
-# ✅ Public repos that work without auth:
 MODEL_CANDIDATES += [
     ("Javvanny/yolov8m_flying_objects_detection", "yolov8m/weights/best.pt"),  # drones/planes/helis/birds
     ("doguilmak/Drone-Detection-YOLOv8x", "weight/best.pt"),                   # drone-focused (larger)
 ]
 
+# =========================
+# LABEL TRANSLATION (RU -> EN)
+# =========================
+LABEL_MAP = {
+    "БПЛА": "UAV",
+    "БПЛА коптер": "Drone",
+    "квадрокоптер": "Quadcopter",
+    "квадроcамолет": "Quadcopter",
+    "самолет": "Airplane",
+    "вертолет": "Helicopter",
+    "птица": "Bird",
+    "человек": "Person",
+    "машина": "Car",
+    "автомобиль": "Car",
+    # add more if your checkpoint uses different strings
+}
+
+def map_label(name: str) -> str:
+    if not isinstance(name, str):
+        return name
+    # exact match first
+    if name in LABEL_MAP:
+        return LABEL_MAP[name]
+    # try case-insensitive
+    low = name.lower()
+    for ru, en in LABEL_MAP.items():
+        if low == ru.lower():
+            return en
+    return name  # fallback
+
+def translate_names_dict(names_dict: Dict[int, str]) -> Dict[int, str]:
+    if not isinstance(names_dict, dict):
+        return names_dict
+    return {k: map_label(v) for k, v in names_dict.items()}
+
 # =========================
 # LAZY STATE
 # =========================
@@ -101,7 +135,8 @@ def _results_to_rows(results) -> List[dict]:
     r = results[0]
     if getattr(r, "boxes", None) is None:
         return rows
-    names = getattr(r, "names", {}) or _model_names or {}
+    names_dict = getattr(r, "names", {}) or _model_names or {}
+    names_dict = translate_names_dict(names_dict)  # translate for table/CSV
     import numpy as np
     xyxy = r.boxes.xyxy.cpu().numpy() if hasattr(r.boxes, "xyxy") else np.zeros((0, 4))
     confs = r.boxes.conf.cpu().numpy() if hasattr(r.boxes, "conf") else np.zeros((0,))
@@ -109,9 +144,9 @@ def _results_to_rows(results) -> List[dict]:
     for i, box in enumerate(xyxy):
         x1, y1, x2, y2 = [float(v) for v in box.tolist()]
         cls_idx = int(clss[i]) if i < len(clss) else -1
-        cls_name = names.get(cls_idx, str(cls_idx))
+        cls_name = names_dict.get(cls_idx, str(cls_idx))
         rows.append({
-            "class": cls_name,
+            "class": map_label(cls_name),
             "confidence": float(confs[i]) if i < len(confs) else None,
             "x1": x1, "y1": y1, "x2": x2, "y2": y2,
             "width": x2 - x1, "height": y2 - y1,
@@ -121,7 +156,8 @@ def _results_to_rows(results) -> List[dict]:
 def _count_by_class(rows: List[dict]) -> Dict[str, int]:
     d: Dict[str, int] = {}
     for r in rows:
-        d[r["class"]] = d.get(r["class"], 0) + 1
+        name = map_label(r.get("class", ""))
+        d[name] = d.get(name, 0) + 1
     return d
 
 def _write_video(base_path: str, fps: float, w: int, h: int):
@@ -173,30 +209,32 @@ def _save_pdf(title: str, summary: str, counts: Dict[str, int], annotated_image_
     c.showPage(); c.save()
     return out_path
 
-# ---- NEW: normalize rows to avoid DataFrame truthiness issues ----
 def _normalize_rows(table_rows):
-    """
-    Accepts pandas.DataFrame OR list[dict] OR list[list] and returns list[dict] with expected keys.
-    """
+    """Accept pandas.DataFrame OR list[dict] OR list[list]; return list[dict]."""
     try:
         import pandas as pd
         if isinstance(table_rows, pd.DataFrame):
             return table_rows.to_dict(orient="records")
     except Exception:
         pass
-
     if isinstance(table_rows, list) and (not table_rows or isinstance(table_rows[0], dict)):
         return table_rows or []
-
     if isinstance(table_rows, list) and table_rows and isinstance(table_rows[0], list):
         headers = ["class","confidence","x1","y1","x2","y2","width","height"]
         return [dict(zip(headers, row)) for row in table_rows]
-
     return []
 
 # =========================
-# INFERENCE (SAFE WRAPPERS)
+# INFERENCE (SAFE WRAPPERS) + ENGLISH OVERLAY
 # =========================
+def _apply_english_overlay(r):
+    """Override r.names with English mapping so r.plot draws English labels."""
+    try:
+        if hasattr(r, "names") and isinstance(r.names, dict):
+            r.names = translate_names_dict(r.names)
+    except Exception:
+        pass
+
 def detect_image_safe(image, conf: float, iou: float):
     try:
         if image is None:
@@ -205,8 +243,9 @@ def detect_image_safe(image, conf: float, iou: float):
         model = _get_model(conf, iou)
         results = model.predict(image, imgsz=960, verbose=False)
         r = results[0]
+        _apply_english_overlay(r)      # <- ensure overlay text is English
         rows = _results_to_rows(results)
-        annotated = r.plot()  # BGR ndarray
+        annotated = r.plot()           # BGR ndarray with English labels
         counts = _count_by_class(rows)
         summary = "Detections: " + (", ".join(f"{k}: {v}" for k, v in counts.items()) if rows else "none")
 
@@ -263,6 +302,7 @@ def detect_video_safe(video_path: str, conf: float, iou: float, max_frames: int
 
                 results = model.predict(frame, imgsz=960, verbose=False)
                 r = results[0]
+                _apply_english_overlay(r)  # <- English overlay per frame
 
                 for row in _results_to_rows(results):
                     totals[row["class"]] = totals.get(row["class"], 0) + 1
@@ -286,6 +326,9 @@ def detect_video_safe(video_path: str, conf: float, iou: float, max_frames: int
 def export_pdf_img(summary: str, table_rows, annotated_tmp_jpg: Optional[str]):
     try:
         rows = _normalize_rows(table_rows)
+        # ensure English in report
+        for r in rows:
+            r["class"] = map_label(r.get("class", ""))
         counts = _count_by_class(rows)
         return _save_pdf(
             "UAV Detector — Image Report",
@@ -301,6 +344,8 @@ def export_pdf_vid(summary: str, counts_json: str):
         counts = json.loads(counts_json) if counts_json else {}
     except Exception:
         counts = {}
+    # map to English just in case
+    counts = {map_label(k): v for k, v in counts.items()}
     try:
         return _save_pdf("UAV Detector — Video Report", summary or "No summary.", counts or {}, None)
     except Exception as e:
@@ -310,8 +355,8 @@ def export_pdf_vid(summary: str, counts_json: str):
 # UI (local embedded samples)
 # =========================
 NOTE = (
-    "UAV model: detects drones (class names vary per checkpoint, e.g., 'drone', 'uav'). "
-    "Ensure the drone is ≥30–40 px on the short side for reliable detection."
+    "UAV model: detects drones (class names vary per checkpoint). "
+    "All labels are translated to English in the UI and exports."
 )
 
 with gr.Blocks(title="UAV / Drone Detector (YOLO)") as demo:
@@ -330,7 +375,6 @@ If they’re missing, you can still upload your own.
         # ---------- IMAGE ----------
         with gr.TabItem("Image"):
             with gr.Row():
-                # Use type="filepath" so embedded path loads directly. Uploads also pass a path.
                 image_in = gr.Image(
                     value=EMBED_IMG if os.path.exists(EMBED_IMG) else None,
                     type="filepath",