Add BCE when no annotations

Correction of focal loss

Author: rvandeghen

efficientdet/loss.py

@@ -52,12 +52,36 @@ def forward(self, classifications, regressions, anchors, annotations, **kwargs):
             bbox_annotation = bbox_annotation[bbox_annotation[:, 4] != -1]
 
             if bbox_annotation.shape[0] == 0:
-                if torch.cuda.is_available():
+                if torch.cuda.is_available()
+                    
+                    targets = torch.zeros_like(classification)
+                    alpha_factor = torch.ones_like(targets) * alpha
+                    targets = targets.cuda()
+                    alpha_factor = alpha_factor.cuda()
+                    alphe_factot = 1. - alpha_factor
+                    focal_weight = classification
+                    focal_weight = alpha_factor * torch.pow(focal_weight, gamma)
+                    
+                    bce = -((1.0 - targets) * torch.log(1.0 - classification))
+                    
+                    cls_loss = focal_weight * bce
+                    
                     regression_losses.append(torch.tensor(0).to(dtype).cuda())
-                    classification_losses.append(torch.tensor(0).to(dtype).cuda())
+                    classification_losses.append(cls_loss.sum(), min=1.0))
                 else:
+                    
+                    targets = torch.zeros_like(classification)
+                    alpha_factor = torch.ones_like(targets) * alpha
+                    alphe_factot = 1. - alpha_factor
+                    focal_weight = classification
+                    focal_weight = alpha_factor * torch.pow(focal_weight, gamma)
+                    
+                    bce = -((1.0 - targets) * torch.log(1.0 - classification))
+                    
+                    cls_loss = focal_weight * bce
+                    
                     regression_losses.append(torch.tensor(0).to(dtype))
-                    classification_losses.append(torch.tensor(0).to(dtype))
+                    classification_losses.append(cls_loss.sum(), min=1.0))
 
                 continue