list literal is faster than using list constructor to create a new list

src/eynollah/utils/__init__.py

@@ -138,8 +138,7 @@ def return_x_start_end_mothers_childs_and_type_of_reading_order(
         min_ys=np.min(y_sep)
         max_ys=np.max(y_sep)
 
-        y_mains=[]
+        y_mains= [min_ys]
-        y_mains.append(min_ys)
         y_mains_sep_ohne_grenzen=[]
 
         for ii in range(len(new_main_sep_y)):
@@ -493,8 +492,7 @@ def find_num_col(regions_without_separators, num_col_classifier, tables, multipl
                 # print(forest[np.argmin(z[forest]) ] )
                 if not isNaN(forest[np.argmin(z[forest])]):
                     peaks_neg_true.append(forest[np.argmin(z[forest])])
-                forest = []
+                forest = [peaks_neg_fin[i + 1]]
-                forest.append(peaks_neg_fin[i + 1])
         if i == (len(peaks_neg_fin) - 1):
             # print(print(forest[np.argmin(z[forest]) ] ))
             if not isNaN(forest[np.argmin(z[forest])]):
@@ -662,8 +660,7 @@ def find_num_col_only_image(regions_without_separators, multiplier=3.8):
                 # print(forest[np.argmin(z[forest]) ] )
                 if not isNaN(forest[np.argmin(z[forest])]):
                     peaks_neg_true.append(forest[np.argmin(z[forest])])
-                forest = []
+                forest = [peaks_neg_fin[i + 1]]
-                forest.append(peaks_neg_fin[i + 1])
         if i == (len(peaks_neg_fin) - 1):
             # print(print(forest[np.argmin(z[forest]) ] ))
             if not isNaN(forest[np.argmin(z[forest])]):
@@ -1235,8 +1232,7 @@ def order_of_regions(textline_mask, contours_main, contours_header, y_ref):
         y_max_header = np.array([np.max(contours_header[j][:, 0, 1]) for j in range(len(contours_header))])
         # print(cy_main,'mainy')
 
-    peaks_neg_new = []
+    peaks_neg_new = [0 + y_ref]
-    peaks_neg_new.append(0 + y_ref)
     for iii in range(len(peaks_neg)):
         peaks_neg_new.append(peaks_neg[iii] + y_ref)
     peaks_neg_new.append(textline_mask.shape[0] + y_ref)
@@ -1404,8 +1400,7 @@ def combine_hor_lines_and_delete_cross_points_and_get_lines_features_back_new(
     return img_p_in[:,:,0], special_separators
 
 def return_points_with_boundies(peaks_neg_fin, first_point, last_point):
-    peaks_neg_tot = []
+    peaks_neg_tot = [first_point]
-    peaks_neg_tot.append(first_point)
     for ii in range(len(peaks_neg_fin)):
         peaks_neg_tot.append(peaks_neg_fin[ii])
     peaks_neg_tot.append(last_point)
@@ -1588,8 +1583,7 @@ def find_number_of_columns_in_document(region_pre_p, num_col_classifier, tables,
     args_cy_splitter=np.argsort(cy_main_splitters)
     cy_main_splitters_sort=cy_main_splitters[args_cy_splitter]
 
-    splitter_y_new=[]
+    splitter_y_new= [0]
-    splitter_y_new.append(0)
     for i in range(len(cy_main_splitters_sort)):
         splitter_y_new.append(  cy_main_splitters_sort[i] )
     splitter_y_new.append(region_pre_p.shape[0])
@@ -1663,8 +1657,7 @@ def return_boxes_of_images_by_order_of_reading_new(
                         num_col, peaks_neg_fin = find_num_col(
                             regions_without_separators[int(splitter_y_new[i]):int(splitter_y_new[i+1]),:],
                             num_col_classifier, tables, multiplier=3.)
-                    peaks_neg_fin_early=[]
+                    peaks_neg_fin_early= [0]
-                    peaks_neg_fin_early.append(0)
                     #print(peaks_neg_fin,'peaks_neg_fin')
                     for p_n in peaks_neg_fin:
                         peaks_neg_fin_early.append(p_n)

src/eynollah/utils/contour.py

@@ -239,8 +239,7 @@ def do_back_rotation_and_get_cnt_back(contour_par, index_r_con, img, slope_first
 
     cont_int, _ = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
     if len(cont_int)==0:
-        cont_int = []
+        cont_int = [contour_par]
-        cont_int.append(contour_par)
         confidence_contour = 0
     else:
         cont_int[0][:, 0, 0] = cont_int[0][:, 0, 0] + np.abs(img_copy.shape[1] - img.shape[1])

src/eynollah/utils/separate_lines.py

@@ -1174,8 +1174,7 @@ def separate_lines_new_inside_tiles(img_path, thetha):
             if diff_peaks[i] > cut_off:
                 if not np.isnan(forest[np.argmin(z[forest])]):
                     peaks_neg_true.append(forest[np.argmin(z[forest])])
-                forest = []
+                forest = [peaks_neg[i + 1]]
-                forest.append(peaks_neg[i + 1])
         if i == (len(peaks_neg) - 1):
             if not np.isnan(forest[np.argmin(z[forest])]):
                 peaks_neg_true.append(forest[np.argmin(z[forest])])
@@ -1195,8 +1194,7 @@ def separate_lines_new_inside_tiles(img_path, thetha):
             if diff_peaks_pos[i] > cut_off:
                 if not np.isnan(forest[np.argmax(z[forest])]):
                     peaks_pos_true.append(forest[np.argmax(z[forest])])
-                forest = []
+                forest = [peaks[i + 1]]
-                forest.append(peaks[i + 1])
         if i == (len(peaks) - 1):
             if not np.isnan(forest[np.argmax(z[forest])]):
                 peaks_pos_true.append(forest[np.argmax(z[forest])])

train/inference.py

@@ -305,8 +305,7 @@ class sbb_predict:
             input_1= np.zeros( (inference_bs, img_height, img_width,3))
 
 
-            starting_list_of_regions = []
+            starting_list_of_regions = [list(range(labels_con.shape[2]))]
-            starting_list_of_regions.append( list(range(labels_con.shape[2])) )
 
             index_update = 0
             index_selected = starting_list_of_regions[0]

train/train.py

@@ -365,8 +365,7 @@ def run(_config, n_classes, n_epochs, input_height,
 
         y_tot=np.zeros((testX.shape[0],n_classes))
 
-        score_best=[]
+        score_best= [0]
-        score_best.append(0)
 
         num_rows = return_number_of_total_training_data(dir_train)
         weights=[]