Upload 4 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+NotoNaskhArabic-VariableFont_wght[[:space:]](1).ttf filter=lfs diff=lfs merge=lfs -text

NotoNaskhArabic-VariableFont_wght (1).ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6614f7a5a98f4397f10149d6c199a5ee7996c48c609c604de30cd9049c5c00c3
+size 305080

app.py

@@ -0,0 +1,263 @@
+import time
+import os
+
+import cv2
+import numpy as np
+import gradio as gr
+from ultralytics import YOLO
+from PIL import Image, ImageDraw, ImageFont
+import arabic_reshaper
+from bidi.algorithm import get_display
+
+import google.generativeai as genai
+
+# ==========================
+# ⚠️ هنا تحط مفتاحك الجديد
+# ==========================
+
+GEMINI_API_KEY = "AIzaSyAvm28ZnTMaZ1Jtg9sYM-EO4qlAN2W4BIQ"
+
+# اضبط الـ API
+genai.configure(api_key=GEMINI_API_KEY)
+
+# ==========================
+# إعدادات YOLO + الثوابت
+# ==========================
+
+WEIGHTS_PATH = "best.pt"
+IMG_SIZE = 256
+CONF_THRESHOLD = 0.5
+
+MIN_STABLE_FRAMES = 3
+WARN_BEFORE_RESET = 1.5
+RESET_DELAY = 2.5
+
+arabic_map = {
+    "aleff": "ا", "bb": "ب", "ta": "ت", "thaa": "ث", "jeem": "ج",
+    "haa": "ح", "khaa": "خ", "dal": "د", "thal": "ذ", "ra": "ر",
+    "zay": "ز", "seen": "س", "sheen": "ش", "saad": "ص", "dhad": "ض",
+    "taa": "ط", "dha": "ظ", "ain": "ع", "ghain": "غ", "fa": "ف",
+    "gaaf": "ق", "kaaf": "ك", "laam": "ل", "la": "لا", "meem": "م",
+    "nun": "ن", "ha": "ه", "waw": "و", "ya": "ي", "yaa": "ي",
+    "toot": "ة", "al": "ال"
+}
+
+# ==========================
+# نص توجيهات Gemini
+# ==========================
+
+SYSTEM_PROMPT = (
+    "أنت مساعد ذكي يستقبل كلمات أو جمل قصيرة قادمة من مترجم لغة "
+    "الإشارة العربية، ودورك أن تعيد صياغتها كنص عربي واضح ومفهوم، "
+    "أو تشرح معناها باختصار إذا كانت كلمة واحدة."
+)
+
+# ==========================
+# دوال رسم العربي
+# ==========================
+
+def draw_arabic_text(img, text, x, y,
+                     font_path="NotoNaskhArabic-VariableFont_wght.ttf",
+                     font_size=24):
+
+    reshaped = arabic_reshaper.reshape(text)
+    bidi_text = get_display(reshaped)
+
+    img_pil = Image.fromarray(img)
+    draw = ImageDraw.Draw(img_pil)
+
+    try:
+        font = ImageFont.truetype(font_path, font_size)
+    except Exception:
+        font = ImageFont.load_default()
+
+    draw.text((x, y), bidi_text, font=font, fill=(0, 0, 0))
+    return np.array(img_pil)
+
+
+def draw_detections(result, frame, names):
+    boxes = result.boxes
+    detected_labels = []
+
+    if boxes is None or len(boxes) == 0:
+        return frame, detected_labels
+
+    for box in boxes:
+        x1, y1, x2, y2 = map(int, box.xyxy[0])
+        cls_id = int(box.cls[0])
+
+        if isinstance(names, dict):
+            eng_label = names.get(cls_id, str(cls_id))
+        else:
+            eng_label = names[cls_id] if cls_id < len(names) else str(cls_id)
+
+        ar_label = arabic_map.get(eng_label, eng_label)
+        detected_labels.append(ar_label)
+
+        cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
+
+        label_bg_y1 = max(0, y1 - 35)
+        label_bg_y2 = y1
+        cv2.rectangle(frame, (x1, label_bg_y1), (x1 + 140, label_bg_y2),
+                      (0, 255, 0), -1)
+
+        frame = draw_arabic_text(frame, ar_label, x1 + 5, label_bg_y1 + 5)
+
+    return frame, detected_labels
+
+
+# ==========================
+# تحميل YOLO
+# ==========================
+
+print("🔹 Loading YOLO model...")
+model = YOLO(WEIGHTS_PATH)
+print("📚 Classes:", model.names)
+
+
+# ==========================
+# Gemini API Call
+# ==========================
+
+def call_gemini_on_word(word: str) -> str:
+    if not word:
+        return ""
+
+    try:
+        model_g = genai.GenerativeModel("gemini-1.5-flash")
+
+        prompt = (
+            SYSTEM_PROMPT +
+            f"\n\nالنص القادم من مترجم لغة الإشارة هو: «{word}».\n"
+            "اكتب جملة قصيرة أو شرحًا بسيطًا بالعربية اعتمادًا على هذا النص."
+        )
+
+        response = model_g.generate_content(prompt)
+        return response.text or ""
+    except Exception as e:
+        return f"خطأ Gemini: {e}"
+
+
+# ==========================
+# معالجة الفريم
+# ==========================
+
+def process_frame(frame,
+                  current_word="",
+                  last_label=None,
+                  stable_count=0,
+                  last_letter_time=None,
+                  chat_history=None):
+
+    if chat_history is None:
+        chat_history = []
+
+    frame_bgr = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
+    frame_bgr = cv2.flip(frame_bgr, 1)
+
+    results = model.predict(frame_bgr, conf=CONF_THRESHOLD,
+                            imgsz=IMG_SIZE, verbose=False)[0]
+
+    annotated, labels = draw_detections(results, frame_bgr, model.names)
+
+    if labels:
+        current_label = labels[0]
+        if current_label == last_label:
+            stable_count += 1
+        else:
+            last_label = current_label
+            stable_count = 1
+
+        if stable_count >= MIN_STABLE_FRAMES:
+            current_word += current_label
+            last_letter_time = time.time()
+            stable_count = 0
+
+    status_text = ""
+
+    if current_word and last_letter_time is not None:
+        elapsed = time.time() - last_letter_time
+
+        if elapsed > RESET_DELAY:
+            final_text = current_word
+
+            chat_history.append(["🖐️ من الإشارات", final_text])
+
+            gpt_reply = call_gemini_on_word(final_text)
+
+            if gpt_reply:
+                chat_history.append(["🤖 المساعد", gpt_reply])
+
+            current_word = ""
+            last_label = None
+            stable_count = 0
+            last_letter_time = None
+        elif elapsed > WARN_BEFORE_RESET:
+            status_text = f"الكلمة الحالية: {current_word} (سيتم إنهاؤها قريبًا)"
+        else:
+            status_text = f"الكلمة الحالية: {current_word}"
+
+    annotated_rgb = cv2.cvtColor(annotated, cv2.COLOR_BGR2RGB)
+
+    return (
+        annotated_rgb,
+        status_text,
+        current_word,
+        last_label,
+        stable_count,
+        last_letter_time,
+        chat_history,
+        chat_history,
+    )
+
+
+# ==========================
+# واجهة Gradio
+# ==========================
+
+with gr.Blocks() as demo:
+    gr.Markdown("## ASL → Arabic Chat (YOLO + Gemini)")
+
+    with gr.Row():
+        cam = gr.Image(
+            sources=["webcam"],
+            streaming=True,
+            type="numpy",
+            label="الكاميرا",
+        )
+        video_out = gr.Image(label="النتيجة")
+
+    word_status = gr.Markdown()
+    chatbox = gr.Chatbot(label="الشات (إشارة → نص)")
+
+    state_current_word = gr.State("")
+    state_last_label = gr.State(None)
+    state_stable_count = gr.State(0)
+    state_last_letter_time = gr.State(None)
+    state_chat_history = gr.State([])
+
+    cam.stream(
+        fn=process_frame,
+        inputs=[
+            cam,
+            state_current_word,
+            state_last_label,
+            state_stable_count,
+            state_last_letter_time,
+            state_chat_history,
+        ],
+        outputs=[
+            video_out,
+            word_status,
+            state_current_word,
+            state_last_label,
+            state_stable_count,
+            state_last_letter_time,
+            state_chat_history,
+            chatbox,
+        ],
+    )
+
+
+if __name__ == "__main__":
+    demo.launch()

best (2).pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1126bf72b6b69eb9e608ad6132a9a9411c37854e1b08f5bb6ccbe8f6f0418c0
+size 52045963

requirements.txt

@@ -0,0 +1,7 @@
+ultralytics
+opencv-python-headless
+Pillow
+arabic-reshaper
+python-bidi
+gradio
+google-generativeai