Update!

app.py

@@ -8,18 +8,26 @@ def load_essential_models(scaler_6_path, scaler_8_path, scaler_full_path, clf_6_
     scaler_8 = pickle.load(open(file=scaler_8_path, mode='rb'))
     scaler_21 = pickle.load(open(file=scaler_full_path, mode='rb'))
     clf_6 = pickle.load(open(file=clf_6_path, mode='rb'))
-    clf_8 = load_model(clf_8_path)
-    clf_21 = load_model(clf_full_path)
+    clf_8 = load_model(clf_8_path, compile=False)
+    clf_21 = load_model(clf_full_path, compile=False)
     return scaler_6, scaler_8, scaler_21, clf_6, clf_8, clf_21
 
-scaler_6, scaler_8, scaler_21, clf_6, clf_8, clf_21 = load_essential_models(scaler_6_path='./Models/6/scaler.pkl', scaler_8_path='./Models/8/scaler.pkl', scaler_full_path='./Models/21/scaler.pkl', clf_6_path='./Models/6/svc_fs_tune.pkl', clf_8_path='./Models/8/ANNs_8base.h5', clf_full_path='./Models/21/ANNs_full.h5')
+def convert_prediction(prediction):
+    if prediction == [0]:
+        return 'S'
+    elif prediction == [1]:
+        return 'M'
+    else:
+        return 'L'
+
+scaler_6, scaler_8, scaler_21, clf_6, clf_8, clf_21 = load_essential_models(scaler_6_path='./Models/6/scaler.pkl', scaler_8_path='./Models/8/scaler.pkl', scaler_full_path='./Models/21/scaler.pkl', clf_6_path='./Models/6/svc_fs_tune.pkl', clf_8_path='./Models/8/ANN_8.h5', clf_full_path='./Models/21/ANNs_full.h5')
 
 st.title('Demo System for Vietnamese Woman Bra Size Classifier')
 
 header_col_1, header_col_2 = st.columns([3, 2])
 num_of_features = header_col_1.selectbox(label='Please select the number of measurements you have:', options=['6 measurements', '8 measurements', '21 measurements'])
 
-h, w, bmi, vtn, vn, vcn, cn, cnnp, cnnt, cntp, cntt, ccnp, ccnt, snt, sndp, sndt, xup, xut, cl, ttp, ttt = [0.0]*21
+# h, w, bmi, vtn, vn, vcn, cn, cnnp, cnnt, cntp, cntt, ccnp, ccnt, snt, sndp, sndt, xup, xut, cl, ttp, ttt = [0.0]*21
 
 if num_of_features == '6 measurements':
     sample_options_6 = header_col_2.selectbox(label='Sample options:', options=['Sample 1 (Small)', 'Sample 2 (Medium)', 'Sample 3 (Large)'])
@@ -50,32 +58,32 @@ elif num_of_features == '8 measurements':
     sample_options_8 = header_col_2.selectbox(label='Sample options:', options=['Sample 1 (Small)', 'Sample 2 (Medium)', 'Sample 3 (Large)'])
     container_col_1, container_col_2, container_col_3, container_col_4 = st.columns([1, 1, 1, 1])
     if sample_options_8 == 'Sample 1 (Small)':
-        ttp = container_col_1.number_input(label='Your ttp:', value=325.10, min_value=0.00, step=0.01)
-        cl = container_col_2.number_input(label='Your cl:', value=8.90, min_value=0.00, step=0.01)
-        cnnp = container_col_3.number_input(label='Your cnnp:', value=12.30, min_value=0.00, step=0.01)
-        vn = container_col_4.number_input(label='Your vn:', value=83.20, min_value=0.00, step=0.01)
-        vtn = container_col_1.number_input(label='Your vtn: ', value=80.10, min_value=0.00, step=0.01)
-        ccnt = container_col_2.number_input(label='Your ccnt:', value=11.10, min_value=0.00, step=0.01)
-        cntp = container_col_3.number_input(label='Your cntp:', value=8.70, min_value=0.00, step=0.01)
-        vcn = container_col_4.number_input(label='Your vcn:', value=74.30, min_value=0.00, step=0.01)
+        ttp = container_col_1.number_input(label='Your ttp:', value=287.50, min_value=0.00, step=0.01)
+        cl = container_col_2.number_input(label='Your cl:', value=7.00, min_value=0.00, step=0.01)
+        cnnp = container_col_3.number_input(label='Your cnnp:', value=8.50, min_value=0.00, step=0.01)
+        vn = container_col_4.number_input(label='Your vn:', value=74.50, min_value=0.00, step=0.01)
+        vtn = container_col_1.number_input(label='Your vtn: ', value=74.00, min_value=0.00, step=0.01)
+        cnnt = container_col_2.number_input(label='Your cnnt:', value=8.90, min_value=0.00, step=0.01)
+        cntp = container_col_3.number_input(label='Your cntp:', value=8.50, min_value=0.00, step=0.01)
+        vcn = container_col_4.number_input(label='Your vcn:', value=67.50, min_value=0.00, step=0.01)
     elif sample_options_8 == 'Sample 2 (Medium)':
-        ttp = container_col_1.number_input(label='Your ttp:', value=385.80, min_value=0.00, step=0.01)
-        cl = container_col_2.number_input(label='Your cl:', value=13.60, min_value=0.00, step=0.01)
-        cnnp = container_col_3.number_input(label='Your cnnp:', value=8.80, min_value=0.00, step=0.01)
-        vn = container_col_4.number_input(label='Your vn:', value=83.20, min_value=0.00, step=0.01)
-        vtn = container_col_1.number_input(label='Your vtn: ', value=82.50, min_value=0.00, step=0.01)
-        ccnt = container_col_2.number_input(label='Your ccnt:', value=9.30, min_value=0.00, step=0.01)
-        cntp = container_col_3.number_input(label='Your cntp:', value=9.10, min_value=0.00, step=0.01)
-        vcn = container_col_4.number_input(label='Your vcn:', value=69.60, min_value=0.00, step=0.01)
-    else:
-        ttp = container_col_1.number_input(label='Your ttp:', value=453.30, min_value=0.00, step=0.01)
+        ttp = container_col_1.number_input(label='Your ttp:', value=489.60, min_value=0.00, step=0.01)
         cl = container_col_2.number_input(label='Your cl:', value=13.00, min_value=0.00, step=0.01)
-        cnnp = container_col_3.number_input(label='Your cnnp:', value=12.50, min_value=0.00, step=0.01)
-        vn = container_col_4.number_input(label='Your vn:', value=85.30, min_value=0.00, step=0.01)
-        vtn = container_col_1.number_input(label='Your vtn: ', value=78.50, min_value=0.00, step=0.01)
-        ccnt = container_col_2.number_input(label='Your ccnt:', value=12.40, min_value=0.00, step=0.01)
-        cntp = container_col_3.number_input(label='Your cntp:', value=10.00, min_value=0.00, step=0.01)
-        vcn = container_col_4.number_input(label='Your vcn:', value=72.30, min_value=0.00, step=0.01)
+        cnnp = container_col_3.number_input(label='Your cnnp:', value=12.40, min_value=0.00, step=0.01)
+        vn = container_col_4.number_input(label='Your vn:', value=78.30, min_value=0.00, step=0.01)
+        vtn = container_col_1.number_input(label='Your vtn: ', value=77.20, min_value=0.00, step=0.01)
+        cnnt = container_col_2.number_input(label='Your cnnt:', value=11.60, min_value=0.00, step=0.01)
+        cntp = container_col_3.number_input(label='Your cntp:', value=8.70, min_value=0.00, step=0.01)
+        vcn = container_col_4.number_input(label='Your vcn:', value=65.30, min_value=0.00, step=0.01)
+    else:
+        ttp = container_col_1.number_input(label='Your ttp:', value=568.2, min_value=0.00, step=0.01)
+        cl = container_col_2.number_input(label='Your cl:', value=13.7, min_value=0.00, step=0.01)
+        cnnp = container_col_3.number_input(label='Your cnnp:', value=14.0, min_value=0.00, step=0.01)
+        vn = container_col_4.number_input(label='Your vn:', value=88.7, min_value=0.00, step=0.01)
+        vtn = container_col_1.number_input(label='Your vtn: ', value=86.5, min_value=0.00, step=0.01)
+        cnnt = container_col_2.number_input(label='Your cnnt:', value=12.5, min_value=0.00, step=0.01)
+        cntp = container_col_3.number_input(label='Your cntp:', value=10.7, min_value=0.00, step=0.01)
+        vcn = container_col_4.number_input(label='Your vcn:', value=75.0, min_value=0.00, step=0.01)
 else:
     sample_options_21 = header_col_2.selectbox(label='Sample options:', options=['Sample 1 (Small)', 'Sample 2 (Medium)', 'Sample 3 (Large)'])
     container_col_1, container_col_2, container_col_3, container_col_4, container_col_5 = st.columns([1, 1, 1, 1, 1])
@@ -153,17 +161,17 @@ with col_3:
 
 if predict:
     if num_of_features == '6 measurements':
-        X_6 = [[vtn, vn, vcn, cl, ttp, ttt]]
+        X_6 = np.array([[vtn, vn, vcn, cl, ttp, ttt]])
         X_6 = scaler_6.transform(X_6)
         y_6 = clf_6.predict(X_6)
-        st.success(f'We recommend you choosing {y_6} size!')
+        st.success(f'We recommend you choosing {convert_prediction(y_6)} size!')
     elif num_of_features == '8 measurements':
-        X_8 = [[ttp, cl, cnnp, vn, vtn, cnnt, cntp, vcn]]
+        X_8 = np.array([[ttp, cl, cnnp, vn, vtn, cnnt, cntp, vcn]])
         X_8 = scaler_8.transform(X_8)
         y_8 = clf_8.predict(X_8)
-        st.success(f'We recommend you choosing {np.argmax(y_8, axis=1)} size!')
+        st.success(f'We recommend you choosing {convert_prediction(np.argmax(y_8, axis=1))} size!')
     else:
-        X_21 = [[h, w, bmi, vtn, vn, vcn, cn, cnnp, cnnt, cntp, cntt, ccnp, ccnt, snt, sndp, sndt, xup, xut, cl, ttp, ttt]]
+        X_21 = np.array([[h, w, bmi, vtn, vn, vcn, cn, cnnp, cnnt, cntp, cntt, ccnp, ccnt, snt, sndp, sndt, xup, xut, cl, ttp, ttt]])
         X_21 = scaler_21.transform(X_21)
         y_21 = clf_21.predict(X_21)
-        st.success(f'We recommend you choosing {np.argmax(y_21, axis=1)} size!')
+        st.success(f'We recommend you choosing {convert_prediction(np.argmax(y_21, axis=1))} size!')