fix a bug on postprocessing

Author: WenmuZhou

Diff for: base/base_trainer.py

@@ -53,7 +53,7 @@ def __init__(self, config, model, criterion):
             self.device = torch.device("cpu")
         self.logger_info('train with device {} and pytorch {}'.format(self.device, torch.__version__))
         # metrics
-        self.metrics = {'recall': 0, 'precision': 0, 'hmean': 0, 'train_loss': float('inf'), 'best_model': ''}
+        self.metrics = {'recall': 0, 'precision': 0, 'hmean': 0, 'train_loss': float('inf')}
 
         self.optimizer = self._initialize('optimizer', torch.optim, model.parameters())
 

Diff for: eval.sh

@@ -1 +1 @@
-CUDA_VISIBLE_DEVICES=0 python3 tools/eval.py --model_path '' --img_folder '' --gt_folder '' --save_folder ''
+CUDA_VISIBLE_DEVICES=0 python3 tools/eval.py --model_path ''

Diff for: post_processing/seg_detector_representer.py

@@ -66,7 +66,7 @@ def polygons_from_bitmap(self, pred, _bitmap, dest_width, dest_height):
             # _, sside = self.get_mini_boxes(contour)
             # if sside < self.min_size:
             #     continue
-            score = self.box_score_fast(pred, points.reshape(-1, 2))
+            score = self.box_score_fast(pred, contour)
             if self.box_thresh > score:
                 continue
 
@@ -112,7 +112,7 @@ def boxes_from_bitmap(self, pred, _bitmap, dest_width, dest_height):
             if sside < self.min_size:
                 continue
             points = np.array(points)
-            score = self.box_score_fast(pred, points.reshape(-1, 2))
+            score = self.box_score_fast(pred, contour)
             if self.box_thresh > score:
                 continue
 

Diff for: trainer/trainer.py

@@ -149,8 +149,8 @@ def _on_epoch_finish(self):
             self.epoch_result['lr']))
         net_save_path = '{}/model_latest.pth'.format(self.checkpoint_dir)
 
-        save_best = False
         if self.config['local_rank'] == 0:
+            save_best = False
             if self.config['trainer']['metrics'] == 'hmean':  # 使用f1作为最优模型指标
                 recall, precision, hmean = self._eval(self.epoch_result['epoch'])
 
@@ -166,12 +166,10 @@ def _on_epoch_finish(self):
                     self.metrics['hmean'] = hmean
                     self.metrics['precision'] = precision
                     self.metrics['recall'] = recall
-                    self.metrics['best_model'] = net_save_path
             else:
                 if self.epoch_result['train_loss'] < self.metrics['train_loss']:
                     save_best = True
                     self.metrics['train_loss'] = self.epoch_result['train_loss']
-                    self.metrics['best_model'] = net_save_path
             self._save_checkpoint(self.epoch_result['epoch'], net_save_path, save_best)
 
     def _on_train_finish(self):