Merge pull request #192 from YongWookHa/master

fix: deprecated 'size_average'

Author: Tianxiaomo

tool/darknet2pytorch.py

@@ -308,11 +308,11 @@ def create_network(self, blocks):
                 models.append(model)
             elif block['type'] == 'cost':
                 if block['_type'] == 'sse':
-                    model = nn.MSELoss(size_average=True)
+                    model = nn.MSELoss(reduction='mean')
                 elif block['_type'] == 'L1':
-                    model = nn.L1Loss(size_average=True)
+                    model = nn.L1Loss(reduction='mean')
                 elif block['_type'] == 'smooth':
-                    model = nn.SmoothL1Loss(size_average=True)
+                    model = nn.SmoothL1Loss(reduction='mean')
                 out_filters.append(1)
                 out_strides.append(prev_stride)
                 models.append(model)

tool/region_loss.py

@@ -174,12 +174,12 @@ def forward(self, output, target):
 
         t3 = time.time()
 
-        loss_x = self.coord_scale * nn.MSELoss(size_average=False)(x * coord_mask, tx * coord_mask) / 2.0
-        loss_y = self.coord_scale * nn.MSELoss(size_average=False)(y * coord_mask, ty * coord_mask) / 2.0
-        loss_w = self.coord_scale * nn.MSELoss(size_average=False)(w * coord_mask, tw * coord_mask) / 2.0
-        loss_h = self.coord_scale * nn.MSELoss(size_average=False)(h * coord_mask, th * coord_mask) / 2.0
-        loss_conf = nn.MSELoss(size_average=False)(conf * conf_mask, tconf * conf_mask) / 2.0
-        loss_cls = self.class_scale * nn.CrossEntropyLoss(size_average=False)(cls, tcls)
+        loss_x = self.coord_scale * nn.MSELoss(reduction='sum')(x * coord_mask, tx * coord_mask) / 2.0
+        loss_y = self.coord_scale * nn.MSELoss(reduction='sum')(y * coord_mask, ty * coord_mask) / 2.0
+        loss_w = self.coord_scale * nn.MSELoss(reduction='sum')(w * coord_mask, tw * coord_mask) / 2.0
+        loss_h = self.coord_scale * nn.MSELoss(reduction='sum')(h * coord_mask, th * coord_mask) / 2.0
+        loss_conf = nn.MSELoss(reduction='sum')(conf * conf_mask, tconf * conf_mask) / 2.0
+        loss_cls = self.class_scale * nn.CrossEntropyLoss(reduction='sum')(cls, tcls)
         loss = loss_x + loss_y + loss_w + loss_h + loss_conf + loss_cls
         t4 = time.time()
         if False:

train.py

@@ -263,11 +263,11 @@ def forward(self, xin, labels=None):
             target[..., 2:4] *= tgt_scale
 
             loss_xy += F.binary_cross_entropy(input=output[..., :2], target=target[..., :2],
-                                              weight=tgt_scale * tgt_scale, size_average=False)
-            loss_wh += F.mse_loss(input=output[..., 2:4], target=target[..., 2:4], size_average=False) / 2
-            loss_obj += F.binary_cross_entropy(input=output[..., 4], target=target[..., 4], size_average=False)
-            loss_cls += F.binary_cross_entropy(input=output[..., 5:], target=target[..., 5:], size_average=False)
-            loss_l2 += F.mse_loss(input=output, target=target, size_average=False)
+                                              weight=tgt_scale * tgt_scale, reduction='sum')
+            loss_wh += F.mse_loss(input=output[..., 2:4], target=target[..., 2:4], reduction='sum') / 2
+            loss_obj += F.binary_cross_entropy(input=output[..., 4], target=target[..., 4], reduction='sum')
+            loss_cls += F.binary_cross_entropy(input=output[..., 5:], target=target[..., 5:], reduction='sum')
+            loss_l2 += F.mse_loss(input=output, target=target, reduction='sum')
 
         loss = loss_xy + loss_wh + loss_obj + loss_cls
 
@@ -511,7 +511,6 @@ def evaluate(model, data_loader, cfg, device, logger=None, **kwargs):
                 "scores": scores,
                 "labels": labels,
             }
-        
         evaluator_time = time.time()
         coco_evaluator.update(res)
         evaluator_time = time.time() - evaluator_time