Move lpr_patterns to config (open-edge-platform#209)

Author: AlexanderDokuchaev

tensorflow_toolkit/lpr/chinese_lp/config.py

@@ -21,6 +21,14 @@
 max_lp_length = 20
 rnn_cells_num = 128
 
+# Licens plate patterns
+lpr_patterns = [
+  '^<[^>]*>[A-Z][0-9A-Z]{5}$',
+  '^<[^>]*>[A-Z][0-9A-Z][0-9]{3}<police>$',
+  '^<[^>]*>[A-Z][0-9A-Z]{4}<[^>]*>$',  # <Guangdong>, <Hebei>
+  '^WJ<[^>]*>[0-9]{4}[0-9A-Z]$',
+]
+
 # Path to the folder where all training and evaluation artifacts will be located
 model_dir = os.path.realpath(os.path.join(os.path.dirname(os.path.abspath(__file__)), '..', 'model'))
 if not os.path.exists(model_dir):
@@ -38,7 +46,7 @@ class train:
   opt_type = 'Adam'
 
   save_checkpoints_steps = 1000      # Number of training steps when checkpoint should be saved
-  display_iter = 10
+  display_iter = 100
 
   apply_basic_aug = False
   apply_stn_aug = True

tensorflow_toolkit/lpr/chinese_lp/config_test.py

@@ -21,6 +21,14 @@
 max_lp_length = 20
 rnn_cells_num = 128
 
+# Licens plate patterns
+lpr_patterns = [
+  '^<[^>]*>[A-Z][0-9A-Z]{5}$',
+  '^<[^>]*>[A-Z][0-9A-Z][0-9]{3}<police>$',
+  '^<[^>]*>[A-Z][0-9A-Z]{4}<[^>]*>$',  # <Guangdong>, <Hebei>
+  '^WJ<[^>]*>[0-9]{4}[0-9A-Z]$',
+]
+
 # Path to the folder where all training and evaluation artifacts will be located
 model_dir = os.path.realpath(os.path.join(os.path.dirname(os.path.abspath(__file__)), '..', 'model'))
 if not os.path.exists(model_dir):
@@ -37,7 +45,7 @@ class train:
   grad_noise_scale = 0.001
   opt_type = 'Adam'
 
-  save_checkpoints_steps = 1000      # Number of training steps when checkpoint should be saved
+  save_checkpoints_steps = 500      # Number of training steps when checkpoint should be saved
   display_iter = 10
 
   apply_basic_aug = False

tensorflow_toolkit/lpr/lpr/utils.py

@@ -23,27 +23,12 @@ def dataset_size(fname):
       count += 1
   return count
 
-
-LPR_PATTERNS = [
-  '^<[^>]*>[A-Z][0-9A-Z]{5}$',
-  '^<[^>]*>[A-Z][0-9A-Z][0-9]{3}<police>$',
-  '^<[^>]*>[A-Z][0-9A-Z]{4}<[^>]*>$',  # <Guangdong>, <Hebei>
-  '^WJ<[^>]*>[0-9]{4}[0-9A-Z]$',
-]
-
-def lpr_pattern_check(label):
-  for pattern in LPR_PATTERNS:
+def lpr_pattern_check(label, lpr_patterns):
+  for pattern in lpr_patterns:
     if re.match(pattern, label):
       return True
   return False
 
-def find_best(predictions):
-  for prediction in predictions:
-    if lpr_pattern_check(prediction):
-      return prediction
-  return predictions[0]  # fallback
-
-
 def edit_distance(string1, string2):
   len1 = len(string1) + 1
   len2 = len(string2) + 1
@@ -60,13 +45,13 @@ def edit_distance(string1, string2):
   return tbl[i, j]
 
 
-def accuracy(label, val, vocab, r_vocab):
+def accuracy(label, val, vocab, r_vocab, lpr_patterns):
   pred = decode_beams(val, r_vocab)
   label_len = len(label)
   acc, acc1 = 0, 0
   num = 0
   for i in range(label_len):
-    if not lpr_pattern_check(label[i].decode('utf-8')):  # GT label fails
+    if not lpr_pattern_check(label[i].decode('utf-8'), lpr_patterns):  # GT label fails
       print('GT label fails: ' + label[i].decode('utf-8'))
       continue
     best = pred[i]

tensorflow_toolkit/lpr/tools/eval.py

@@ -131,7 +131,7 @@ def validate(config):
       num = 0
       for _ in range(steps):
         val, slabel, _ = sess.run([d_predictions, label_val, file_names])
-        acc, acc1, num_ = accuracy(slabel, val, config.vocab, config.r_vocab)
+        acc, acc1, num_ = accuracy(slabel, val, config.vocab, config.r_vocab, config.lpr_patterns)
         mean_accuracy += acc
         mean_accuracy_minus_1 += acc1
         num += num_