Initial commit

docbank.py

@@ -172,7 +172,7 @@ class Docbank(datasets.GeneratorBasedBuilder):
         data_dir = dl_manager.download_and_extract(urls)
         #print(data_dir)
         train_txts = glob(data_dir['train'] + '/train/txt/*.csv')
-        print(train_txts[0])
+        #print(train_txts[0])
         train_data = [(txt, data_dir['train'] + '/train/img/' + os.path.basename(txt)[:-4] + '.jpg') for txt in train_txts]
         test_txts = glob(data_dir['test'] + '/test/txt/*.csv')
         test_data = [(txt, data_dir['test'] + '/test/img/' + os.path.basename(txt)[:-4] + '.jpg') for txt in test_txts]
@@ -222,7 +222,7 @@ class Docbank(datasets.GeneratorBasedBuilder):
         # print(filepath)
         key = 0
         for f_fp_txt, f_fp_img in filepath:
-            print(f_fp_txt)
+            #print(f_fp_txt)
             f_id = key
             #f_fp_txt = f['filepath_txt']
             #f_fp_img = f['filepath_img']
@@ -238,9 +238,9 @@ class Docbank(datasets.GeneratorBasedBuilder):
 
             try:
                 with open(f_fp_txt, encoding='utf-8') as csvfile:
-                    reader = csv.reader(csvfile, delimiter=',')
+                    reader = csv.DictReader(csvfile, delimiter=',')
                     for row in reader:
-                        print(row)
+                        #print(row)
                         # normalized_bbox = normalize_bbox(row[1:5], size)
                         normalized_bbox = [int(x) for x in row[1:5]]
                         tokens.append(row[0])