modify dataset

@@ -239,7 +239,7 @@ def build_dataset(img_dir, go_res_dir, label_dir, top_text_list, skip_list, save
                         [label_bbox[0], label_bbox[3]],
                     ]
                     iou = bbox_iou(go_bbox_rebuild, label_bbox_rebuild)
-                    if iou >= 0.4:
+                    if iou >= 0.2:
                         label_idx_dict[go_idx] = label_idx 
         
         X = list()
@@ -264,7 +264,7 @@ def build_dataset(img_dir, go_res_dir, label_dir, top_text_list, skip_list, save
                 (x0, y0, x1, y1, x2, y2, x3, y3), text = go_res_list[i]
                 feature_vec = [1.]
                 feature_vec.extend(simple_word2vec(text))
-                feature_vec.extend([x0/w, y0/h, x1/w, y1/h, x2/w, y2/h, x3/w, y3/h])
+                feature_vec.extend([(x0/w)*2-1, (y0/h)*2-1, (x1/w)*2-1, (y1/h)*2-1, (x2/w)*2-1, (y2/h)*2-1, (x3/w)*2-1, (y3/h)*2-1])
                 # feature_vec.extend(jwq_word2vec(text, text_vec_max_lens))
                 feature_vec.extend(jieba_and_tencent_word2vec(text, max_jieba_char))
                 X.append(feature_vec)
@@ -273,9 +273,9 @@ def build_dataset(img_dir, go_res_dir, label_dir, top_text_list, skip_list, save
 
             elif i in label_idx_dict:
                 (x0, y0, x1, y1, x2, y2, x3, y3), text = go_res_list[i]
-                feature_vec = [0.]
+                feature_vec = [-1.]
                 feature_vec.extend(simple_word2vec(text))
-                feature_vec.extend([x0/w, y0/h, x1/w, y1/h, x2/w, y2/h, x3/w, y3/h])
+                feature_vec.extend([(x0/w)*2-1, (y0/h)*2-1, (x1/w)*2-1, (y1/h)*2-1, (x2/w)*2-1, (y2/h)*2-1, (x3/w)*2-1, (y3/h)*2-1])
                 # feature_vec.extend(jwq_word2vec(text, text_vec_max_lens))
                 feature_vec.extend(jieba_and_tencent_word2vec(text, max_jieba_char))
                 X.append(feature_vec)
@@ -285,9 +285,9 @@ def build_dataset(img_dir, go_res_dir, label_dir, top_text_list, skip_list, save
                 y_true.append(base_label_list)
             else:
                 (x0, y0, x1, y1, x2, y2, x3, y3), text = go_res_list[i]
-                feature_vec = [0.]
+                feature_vec = [-1.]
                 feature_vec.extend(simple_word2vec(text))
-                feature_vec.extend([x0/w, y0/h, x1/w, y1/h, x2/w, y2/h, x3/w, y3/h])
+                feature_vec.extend([(x0/w)*2-1, (y0/h)*2-1, (x1/w)*2-1, (y1/h)*2-1, (x2/w)*2-1, (y2/h)*2-1, (x3/w)*2-1, (y3/h)*2-1])
                 # feature_vec.extend(jwq_word2vec(text, text_vec_max_lens))
                 feature_vec.extend(jieba_and_tencent_word2vec(text, max_jieba_char))
                 X.append(feature_vec)

@@ -27,12 +27,12 @@ def simple_word2vec(text):
         else:
             other_num += 1
     
-    vec = [text_len/100,
-           cn_num/text_len,
-           en_num/text_len,
-           digit_num/text_len,
+    vec = [(text_len/100)*2 - 1,
+           (cn_num/text_len)*2 - 1,
+           (en_num/text_len)*2 - 1,
+           (digit_num/text_len)*2 - 1,
            # space_num/text_len,
-           other_num/text_len,
+           (other_num/text_len)*2 - 1,
            ]
 
     # print(text)