Deploy RC column predictor with Gradio

app.py

@@ -0,0 +1,89 @@
+import gradio as gr
+import numpy as np
+import joblib
+import tensorflow as tf
+from huggingface_hub import hf_hub_download
+
+# 1) 허브에서 모델·스케일러 파일 다운로드
+model_path = hf_hub_download(repo_id="dklee2013/rc-column-predictor",
+                             filename="model0526.keras")
+sx_path    = hf_hub_download(repo_id="dklee2013/rc-column-predictor",
+                             filename="sx.save")
+sy_path    = hf_hub_download(repo_id="dklee2013/rc-column-predictor",
+                             filename="sy.save")
+
+model    = tf.keras.models.load_model(model_path)
+scaler_X = joblib.load(sx_path)
+scaler_y = joblib.load(sy_path)
+
+def predict(f_ck, f_yk, N_Ed, M_Edz, M_Edy, cover, rebar_dia, dc_ratio):
+    # 1차 예측
+    X1 = np.array([f_ck, f_yk, N_Ed, M_Edz, M_Edy, cover, rebar_dia]).reshape(1,-1)
+    Y1 = scaler_y.inverse_transform(model.predict(scaler_X.transform(X1)))
+    a  = Y1[0,5]
+
+    # 모멘트 보정
+    M_Edz_sc = M_Edz * (1/dc_ratio)**a
+    M_Edy_sc = M_Edy * (1/dc_ratio)**a
+
+    # 2차 예측
+    X2 = np.array([f_ck, f_yk, N_Ed, M_Edz_sc, M_Edy_sc, cover, rebar_dia]).reshape(1,-1)
+    Y2 = scaler_y.inverse_transform(model.predict(scaler_X.transform(X2)))
+
+    # 결과
+    b_pred = int(round(Y2[0,0]/50)*50)
+    h_pred = int(round(Y2[0,1]/50)*50)
+    rr     = float(Y2[0,2])
+    rx     = int(round(Y2[0,3]))
+    ry     = int(round(Y2[0,4]))
+    z1_a   = float(Y2[0,5])
+    z2_N   = float(Y2[0,6])
+    z3_Mz  = float(Y2[0,7])
+    z4_My  = float(Y2[0,8])
+    SF     = (M_Edz / z3_Mz)**z1_a + (M_Edy / z4_My)**z1_a
+
+    return {
+      "b_pred": b_pred,
+      "h_pred": h_pred,
+      "reinforcement_ratio": rr,
+      "rebar_x": rx,
+      "rebar_y": ry,
+      "exponent_a": round(z1_a,4),
+      "N_Rd": round(z2_N,2),
+      "M_Rdz": round(z3_Mz,2),
+      "M_Rdy": round(z4_My,2),
+      "SF": round(SF,4)
+    }
+
+# 3) Gradio UI 정의
+inputs = [
+  gr.Number(label="f_ck (MPa)", value=40),
+  gr.Number(label="f_yk (MPa)", value=500),
+  gr.Number(label="N_Ed (kN)", value=500),
+  gr.Number(label="M_Edz (kN·m)", value=100),
+  gr.Number(label="M_Edy (kN·m)", value=150),
+  gr.Number(label="cover (mm)", value=40),
+  gr.Number(label="rebar_dia (mm)", value=25),
+  gr.Number(label="dc_ratio", value=0.9, precision=2)
+]
+outputs = {
+  "b_pred": gr.Number(label="b_pred (mm)"),
+  "h_pred": gr.Number(label="h_pred (mm)"),
+  "reinforcement_ratio": gr.Number(label="reinforcement_ratio"),
+  "rebar_x": gr.Number(label="rebar_x (개)"),
+  "rebar_y": gr.Number(label="rebar_y (개)"),
+  "exponent_a": gr.Number(label="exponent a"),
+  "N_Rd": gr.Number(label="N_Rd"),
+  "M_Rdz": gr.Number(label="M_Rdz"),
+  "M_Rdy": gr.Number(label="M_Rdy"),
+  "SF": gr.Number(label="Safety Factor")
+}
+
+demo = gr.Interface(fn=predict,
+                    inputs=inputs,
+                    outputs=outputs,
+                    title="🔧 RC 기둥 단면 예측기",
+                    description="Eurocode 2 기반 ML 모델을 이용한 RC 기둥 단면 예측")
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio
+tensorflow
+scikit-learn
+joblib
+huggingface-hub