Update app.py

app.py

@@ -908,6 +908,192 @@ def apply_brightness(img):
     return Image.fromarray(arr)
 
 
+
+
+# ═══════════════════════════════════════════════════════════════
+# 🧠 AI Image Enhancement Engine
+# Real-ESRGAN + GPU Neural + OpenCV Advanced
+# ═══════════════════════════════════════════════════════════════
+
+ESRGAN_MODEL = None
+AI_ENHANCE_READY = False
+
+def load_ai_enhance_models():
+    global ESRGAN_MODEL, AI_ENHANCE_READY
+    if AI_ENHANCE_READY:
+        return True
+    try:
+        from basicsr.archs.rrdbnet_arch import RRDBNet
+        from realesrgan import RealESRGANer
+        model_path = None
+        for p in ["RealESRGAN_x2plus.pth", "weights/RealESRGAN_x2plus.pth", "/app/RealESRGAN_x2plus.pth"]:
+            if os.path.exists(p):
+                model_path = p; break
+        if model_path is None:
+            try:
+                from huggingface_hub import hf_hub_download
+                model_path = hf_hub_download(repo_id="ai-forever/Real-ESRGAN", filename="RealESRGAN_x2.pth")
+                print("📥 Downloaded Real-ESRGAN from HF Hub")
+            except:
+                try:
+                    model_path = hf_hub_download(repo_id="sberbank-ai/Real-ESRGAN", filename="RealESRGAN_x2.pth")
+                except Exception as e:
+                    print(f"⚠️ Real-ESRGAN download failed: {e}")
+        if model_path:
+            esrgan_net = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=2)
+            ESRGAN_MODEL = RealESRGANer(scale=2, model_path=model_path, model=esrgan_net,
+                tile=384, tile_pad=10, pre_pad=0, half=True if USE_GPU else False,
+                device=DEVICE if torch is not None else 'cpu')
+            print("✅ Real-ESRGAN loaded (x2, tile=384)")
+    except ImportError:
+        print("⚠️ Real-ESRGAN not installed (pip install realesrgan basicsr)")
+    except Exception as e:
+        print(f"⚠️ Real-ESRGAN error: {e}")
+    AI_ENHANCE_READY = True
+    return ESRGAN_MODEL is not None
+
+def ai_enhance_esrgan(img, scale=2):
+    if img is None: return None
+    load_ai_enhance_models()
+    if ESRGAN_MODEL is None: return None
+    try:
+        import cv2
+        img_bgr = cv2.cvtColor(np.array(img.convert('RGB')), cv2.COLOR_RGB2BGR)
+        output, _ = ESRGAN_MODEL.enhance(img_bgr, outscale=scale)
+        return Image.fromarray(cv2.cvtColor(output, cv2.COLOR_BGR2RGB))
+    except Exception as e:
+        print(f"⚠️ ESRGAN: {e}"); return None
+
+def ai_enhance_clahe_adv(img):
+    if img is None: return None
+    try:
+        import cv2
+        arr = np.array(img.convert('RGB'))
+        lab = cv2.cvtColor(arr, cv2.COLOR_RGB2LAB)
+        l, a, b = cv2.split(lab)
+        l = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8,8)).apply(l)
+        return Image.fromarray(cv2.cvtColor(cv2.merge([l,a,b]), cv2.COLOR_LAB2RGB))
+    except Exception as e:
+        print(f"⚠️ CLAHE: {e}"); return None
+
+def ai_enhance_histeq(img):
+    if img is None: return None
+    try:
+        import cv2
+        arr = np.array(img.convert('RGB'))
+        ycrcb = cv2.cvtColor(arr, cv2.COLOR_RGB2YCrCb)
+        ycrcb[:,:,0] = cv2.equalizeHist(ycrcb[:,:,0])
+        return Image.fromarray(cv2.cvtColor(ycrcb, cv2.COLOR_YCrCb2RGB))
+    except Exception as e:
+        print(f"⚠️ HistEQ: {e}"); return None
+
+def ai_enhance_sharpen(img):
+    if img is None: return None
+    try:
+        import cv2
+        arr = np.array(img.convert('RGB'))
+        denoised = cv2.fastNlMeansDenoisingColored(arr, None, 8, 8, 7, 21)
+        kernel = np.array([[-1,-1,-1],[-1,9.5,-1],[-1,-1,-1]])
+        return Image.fromarray(cv2.filter2D(denoised, -1, kernel))
+    except Exception as e:
+        print(f"⚠️ Sharpen: {e}"); return None
+
+def ai_enhance_combined(img):
+    if img is None: return None
+    try:
+        import cv2
+        arr = np.array(img.convert('RGB'))
+        denoised = cv2.fastNlMeansDenoisingColored(arr, None, 6, 6, 7, 21)
+        lab = cv2.cvtColor(denoised, cv2.COLOR_RGB2LAB)
+        l, a, b = cv2.split(lab)
+        l = cv2.createCLAHE(clipLimit=2.5, tileGridSize=(8,8)).apply(l)
+        enhanced = cv2.cvtColor(cv2.merge([l,a,b]), cv2.COLOR_LAB2RGB)
+        kernel = np.array([[0,-0.5,0],[-0.5,3,-0.5],[0,-0.5,0]])
+        return Image.fromarray(np.clip(cv2.filter2D(enhanced, -1, kernel), 0, 255).astype(np.uint8))
+    except Exception as e:
+        print(f"⚠️ Combined: {e}"); return None
+
+def ai_enhance_gpu_neural(img):
+    if img is None or torch is None: return None
+    try:
+        arr = np.array(img.convert('RGB')).astype(np.float32) / 255.0
+        t = torch.from_numpy(arr.transpose(2,0,1)).unsqueeze(0).to(DEVICE)
+        enhanced = t.clone()
+        for ks in [3, 5, 9]:
+            pad = ks // 2
+            blurred = torch.nn.functional.avg_pool2d(
+                torch.nn.functional.pad(t, (pad,pad,pad,pad), mode='reflect'), ks, 1)
+            enhanced = enhanced + 0.3 * (t - blurred)
+        enhanced = torch.clamp(enhanced, 0, 1)
+        result = enhanced.squeeze(0).cpu().numpy().transpose(1,2,0)
+        return Image.fromarray((result * 255).clip(0,255).astype(np.uint8))
+    except Exception as e:
+        print(f"⚠️ GPU Neural: {e}"); return None
+
+def ai_enhance_all(img):
+    if img is None:
+        empty = "<div style='text-align:center;padding:60px;color:#999;direction:rtl;'>"
+        empty += "<div style='font-size:48px;'>🧠</div>"
+        empty += "<div>ارفع صورة سونار للتحسين بالذكاء الاصطناعي</div></div>"
+        return empty, None, None, None, None, None, None
+    start = time_module.time()
+    w_orig, h_orig = img.size
+    results = {}; timings = {}
+    techniques = [
+        ("clahe", "🔆 CLAHE Advanced", ai_enhance_clahe_adv),
+        ("histeq", "📊 Histogram EQ", ai_enhance_histeq),
+        ("sharpen", "⚡ Sharpen + Denoise", ai_enhance_sharpen),
+        ("combined", "🏆 Combined Best", ai_enhance_combined),
+        ("esrgan", "🔬 Real-ESRGAN x2", ai_enhance_esrgan),
+        ("neural", "🌊 GPU Neural", ai_enhance_gpu_neural),
+    ]
+    for key, name, func in techniques:
+        t0 = time_module.time()
+        try:
+            result = func(img)
+            if result is not None:
+                results[key] = result; timings[key] = time_module.time() - t0
+            else: timings[key] = -1
+        except Exception as e:
+            print(f"⚠️ {key}: {e}"); timings[key] = -1
+    elapsed = time_module.time() - start
+    n_ok = sum(1 for v in timings.values() if v >= 0)
+    html = "<div style='font-family:Tajawal,sans-serif;direction:rtl;'>"
+    html += "<div style='background:linear-gradient(135deg,#1a237e,#4A148C);padding:14px;border-radius:12px;text-align:center;color:white;margin-bottom:10px;'>"
+    html += f"<div style='font-size:16px;font-weight:bold;'>🧠 تحسين بالذكاء الاصطناعي — {n_ok}/{len(techniques)} تقنيات نجحت</div>"
+    html += f"<div style='font-size:12px;opacity:0.8;'>الصورة: {w_orig}×{h_orig} | الوقت: {elapsed:.1f}s | {'✅ ESRGAN' if 'esrgan' in results else '❌ ESRGAN'} | {'✅ GPU' if 'neural' in results else '❌ GPU'}</div>"
+    html += "</div>"
+    html += "<div style='display:grid;grid-template-columns:repeat(3,1fr);gap:8px;margin:10px 0;'>"
+    labels = {"clahe":("🔆 CLAHE","#00897B"), "histeq":("📊 HistEQ","#1565C0"), "sharpen":("⚡ Sharpen","#E65100"),
+              "combined":("🏆 Combined","#2E7D32"), "esrgan":("🔬 ESRGAN","#6A1B9A"), "neural":("🌊 Neural","#C62828")}
+    for key, (name, color) in labels.items():
+        t = timings.get(key, -1)
+        if t >= 0:
+            rw, rh = results[key].size
+            html += f"<div style='background:{color}15;border:2px solid {color}40;border-radius:10px;padding:10px;text-align:center;'>"
+            html += f"<div style='font-weight:bold;color:{color};'>{name}</div>"
+            html += f"<div style='font-size:11px;color:#666;'>{rw}×{rh} | ⏱{t:.1f}s</div></div>"
+        else:
+            html += f"<div style='background:#f5f5f5;border:2px solid #ddd;border-radius:10px;padding:10px;text-align:center;'>"
+            html += f"<div style='color:#999;'>{name} ❌</div></div>"
+    html += "</div>"
+    html += "<div style='background:#E3F2FD;padding:10px;border-radius:8px;font-size:12px;margin-top:8px;border:1px solid #90CAF9;'>"
+    html += "<b>💡</b> CLAHE+Combined سريعة وفعالة | Real-ESRGAN أفضل تكبير | GPU Neural تحسين عصبي</div></div>"
+    return (html, results.get("clahe"), results.get("histeq"), results.get("sharpen"),
+            results.get("combined"), results.get("esrgan"), results.get("neural"))
+
+def ai_enhance_single(img, technique):
+    if img is None: return None, "❌ لا توجد صورة"
+    funcs = {"🔆 CLAHE Advanced": ai_enhance_clahe_adv, "📊 Histogram EQ": ai_enhance_histeq,
+             "⚡ Sharpen + Denoise": ai_enhance_sharpen, "🏆 Combined Best": ai_enhance_combined,
+             "🔬 Real-ESRGAN x2": ai_enhance_esrgan, "🌊 GPU Neural": ai_enhance_gpu_neural}
+    func = funcs.get(technique)
+    if func is None: return None, "❌ تقنية غير معروفة"
+    t0 = time_module.time(); result = func(img); elapsed = time_module.time() - t0
+    if result is None: return None, f"❌ {technique} غير متوفرة"
+    rw, rh = result.size
+    return result, f"✅ {technique} | {rw}×{rh} | ⏱ {elapsed:.1f}s"
+
 # ═══════════════════════════════════════════════════════════════
 # 📡 ربط قاعدة البيانات عبر API
 # ══════════════════════════════════════════════════��════════════
@@ -2071,6 +2257,57 @@ with gr.Blocks(title=f"SONAR-AI v{VERSION}") as app:
                 p_btn.click(physics_analyze, inputs=[p_img], outputs=[p_res, p_tbl])
 
 
+
+            # ═══ 🧠 تحسين الصور بالذكاء الاصطناعي ═══
+            with gr.Tab("🧠 تحسين AI"):
+                gr.HTML("""<div style='background:linear-gradient(135deg,#1a237e,#4A148C,#880E4F);padding:14px;border-radius:12px;margin-bottom:10px;color:white;text-align:center;'>
+                    <div style='font-size:18px;font-weight:bold;'>🧠 تحسين صور السونار بالذكاء الاصطناعي</div>
+                    <div style='font-size:12px;opacity:0.8;margin-top:4px;'>Real-ESRGAN · GPU Neural · CLAHE · Histogram EQ · Sharpen+Denoise · Combined</div>
+                </div>""")
+                with gr.Row():
+                    with gr.Column(scale=1):
+                        enh_img = gr.Image(label="📷 صورة السونار الأصلية", type="pil", height=250)
+                        gr.HTML("""<div style='background:#F3E5F5;padding:10px;border-radius:8px;border:1px solid #CE93D8;font-size:11px;margin:6px 0;'>
+                            <b>6 تقنيات:</b><br>
+                            🔆 CLAHE — تحسين تباين محلي (سريع 0.1s)<br>
+                            📊 Histogram EQ — توزيع كثافات (سريع 0.05s)<br>
+                            ⚡ Sharpen+Denoise — حدة + إزالة ضوضاء (0.3s)<br>
+                            🏆 Combined — الكل مجتمع (أفضل نتيجة سريعة)<br>
+                            🔬 Real-ESRGAN — تكبير x2 بشبكة عصبية (GPU 5-15s)<br>
+                            🌊 GPU Neural — تحسين عصبي متعدد المقاييس (GPU 1-3s)
+                        </div>""")
+                        enh_btn_all = gr.Button("🧠 تحسين بكل التقنيات", variant="primary", size="lg")
+                        gr.HTML("<div style='text-align:center;color:#666;font-size:12px;margin:6px 0;'>— أو اختر تقنية واحدة —</div>")
+                        enh_technique = gr.Radio(
+                            choices=["🔆 CLAHE Advanced", "📊 Histogram EQ", "⚡ Sharpen + Denoise",
+                                     "🏆 Combined Best", "🔬 Real-ESRGAN x2", "🌊 GPU Neural"],
+                            value="🏆 Combined Best", label="اختر التقنية")
+                        enh_btn_single = gr.Button("⚡ تحسين سريع", variant="secondary")
+                        enh_single_result = gr.Image(label="النتيجة", type="pil", height=200)
+                        enh_single_status = gr.HTML("")
+                    with gr.Column(scale=2):
+                        enh_html = gr.HTML("<div style='text-align:center;padding:60px;color:#999;direction:rtl;'><div style='font-size:48px;'>🧠</div><div>ارفع صورة سونار للتحسين بالذكاء الاصطناعي</div><div style='font-size:12px;margin-top:8px;'>6 تقنيات — GPU accelerated</div></div>")
+                gr.HTML("<div style='background:#1565C0;padding:8px;border-radius:8px;text-align:center;color:white;font-weight:bold;margin:8px 0;'>📊 مقارنة 6 تقنيات</div>")
+                with gr.Row():
+                    enh_clahe = gr.Image(label="🔆 CLAHE Advanced", type="pil", height=220)
+                    enh_histeq = gr.Image(label="📊 Histogram EQ", type="pil", height=220)
+                    enh_sharpen = gr.Image(label="⚡ Sharpen + Denoise", type="pil", height=220)
+                with gr.Row():
+                    enh_combined = gr.Image(label="🏆 Combined Best", type="pil", height=220)
+                    enh_esrgan = gr.Image(label="🔬 Real-ESRGAN x2", type="pil", height=220)
+                    enh_neural = gr.Image(label="🌊 GPU Neural", type="pil", height=220)
+                gr.HTML("<div style='background:#2E7D32;padding:8px;border-radius:8px;text-align:center;color:white;font-weight:bold;margin:8px 0;'>🔄 مقارنة قبل / بعد</div>")
+                with gr.Row():
+                    enh_before = gr.Image(label="📷 قبل (الأصلية)", type="pil", height=300)
+                    enh_after = gr.Image(label="🏆 بعد (أفضل نتيجة)", type="pil", height=300)
+                def _run_all_slider(img):
+                    html, clahe, histeq, sharpen, combined, esrgan, neural = ai_enhance_all(img)
+                    best = esrgan if esrgan is not None else (combined if combined is not None else img)
+                    return html, clahe, histeq, sharpen, combined, esrgan, neural, img, best
+                enh_btn_all.click(_run_all_slider, inputs=[enh_img],
+                    outputs=[enh_html, enh_clahe, enh_histeq, enh_sharpen, enh_combined, enh_esrgan, enh_neural, enh_before, enh_after])
+                enh_btn_single.click(ai_enhance_single, inputs=[enh_img, enh_technique], outputs=[enh_single_result, enh_single_status])
+
             with gr.Tab("🏷️ قاعدة الأصناف"):
                 gr.HTML(f"<div style='background:#E8F5E9;padding:10px 14px;border-radius:10px;margin-bottom:8px;border-right:4px solid #2E7D32;'><b style='color:#2E7D32;'>📦 {len(CARGO_DATABASE)} صنف</b></div>")
                 cats = gr.Dataframe(value=get_categories())