Upload 3 files

app2.py

@@ -0,0 +1,92 @@
+import gradio as gr
+import cv2
+import numpy as np
+import math
+from ultralytics import YOLO
+
+# تحميل الموديل
+MODEL_PATH = "best.pt"
+model = YOLO(MODEL_PATH, task="pose")
+
+def get_hip_points(image_crop, model, conf=0.25):
+    results = model(image_crop, conf=conf, verbose=False)
+    for r in results:
+        if r.keypoints and r.keypoints.shape[1] >= 2:
+            kpts = r.keypoints.xy[0].cpu().numpy().astype(int)
+            valid_kpts = [p for p in kpts if p[0] != 0 and p[1] != 0]
+            if len(valid_kpts) >= 2:
+                valid_kpts.sort(key=lambda p: p[1])
+                return valid_kpts[-1], valid_kpts[0]
+    return None, None
+
+def calculate_slope_angle(center, rim, other_center):
+    try:
+        if other_center[0] - center[0] == 0: m_h = 0
+        else: m_h = (other_center[1] - center[1]) / (other_center[0] - center[0])
+        if rim[0] - center[0] == 0: m_r = 1e9
+        else: m_r = (rim[1] - center[1]) / (rim[0] - center[0])
+        tan_theta = abs((m_r - m_h) / (1 + m_h * m_r))
+        return math.degrees(math.atan(tan_theta))
+    except: return 0.0
+
+def get_diagnosis_style(angle):
+    # ملاحظة: هون بتقدر تعدل الدرجات لـ 4 حسب مشروعك
+    if angle >= 30: return "Dysplasia (Severity High)", "#FF4B4B"
+    if angle >= 25: return "Borderline / Mild", "#FFA500"
+    return "Normal", "#09AB3B"
+
+def analyze_xray(input_img, conf):
+    image = np.array(input_img)
+    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+    h, w, _ = image.shape
+    mid_x = w // 2
+    
+    right_hip_img = image[:, :mid_x]
+    left_hip_img = image[:, mid_x:]
+    
+    r_tri, r_rim = get_hip_points(right_hip_img, model, conf)
+    l_tri_local, l_rim_local = get_hip_points(left_hip_img, model, conf)
+    
+    if r_tri is None or l_tri_local is None:
+        return image, "Could not detect landmarks. Try lowering confidence."
+
+    l_tri_global = (l_tri_local[0] + mid_x, l_tri_local[1])
+    l_rim_global = (l_rim_local[0] + mid_x, l_rim_local[1])
+    
+    # الرسم (نفس المنطق تبعك)
+    cv2.line(image, tuple(r_tri), tuple(l_tri_global), (0, 255, 255), 3)
+    
+    ang_r = calculate_slope_angle(r_tri, r_rim, l_tri_global)
+    diag_r, color_r = get_diagnosis_style(ang_r)
+    
+    ang_l = calculate_slope_angle(l_tri_global, l_rim_global, r_tri)
+    diag_l, color_l = get_diagnosis_style(ang_l)
+    
+    # إضافة النصوص على الصورة أو إرجاعها كـ Markdown
+    result_text = f"## 🦴 Diagnosis Results:\n"
+    result_text += f"**Right Hip:** {ang_r:.1f}° - {diag_r}\n\n"
+    result_text += f"**Left Hip:** {ang_l:.1f}° - {diag_l}"
+
+    # رسم النقاط والخطوط
+    cv2.line(image, tuple(r_tri), tuple(r_rim), (0, 0, 255), 3)
+    cv2.line(image, tuple(l_tri_global), tuple(l_rim_global), (0, 0, 255), 3)
+    
+    return cv2.cvtColor(image, cv2.COLOR_BGR2RGB), result_text
+
+# واجهة Gradio
+with gr.Blocks(title="DDH AI Expert") as demo:
+    gr.Markdown("# 🏥 DDH AI Expert System")
+    gr.Markdown("قم برفع صورة X-ray لحوض الطفل للكشف التلقائي عن خلع الولادة.")
+    
+    with gr.Row():
+        with gr.Column():
+            input_img = gr.Image(type="pil", label="Upload X-ray")
+            conf_slider = gr.Slider(0.1, 1.0, value=0.25, label="Confidence Threshold")
+            btn = gr.Button("Analyze Image")
+        with gr.Column():
+            output_img = gr.Image(label="Processed Image")
+            output_text = gr.Markdown()
+
+    btn.click(analyze_xray, inputs=[input_img, conf_slider], outputs=[output_img, output_text])
+
+demo.launch()

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bacad1fd22999b106284f1aace0db464287ed72e6e9fad646696d1a1945030f0
+size 6422616

requirements.txt

@@ -0,0 +1,5 @@
+flask==3.0.0
+opencv-python==4.8.1.78
+numpy==2.1.2
+ultralytics==8.0.196
+Pillow==10.1.0