updates

train.py

@@ -53,12 +53,12 @@ def train(
     yl = get_yolo_layers(model)  # yolo layers
     nf = int(model.module_defs[yl[0] - 1]['filters'])  # yolo layer size (i.e. 255)
 
-    if resume:  # Load previously saved PyTorch model
+    if resume:  # Load previously saved model
         if transfer:  # Transfer learning
             chkpt = torch.load(weights + 'yolov3.pt', map_location=device)
-            model.load_state_dict(
-                {k: v for k, v in chkpt['model'].items() if (int(k.split('.')[1]) + 1) not in yl}, strict=False)
-            for (name, p) in model.named_parameters():
+            model.load_state_dict({k: v for k, v in chkpt['model'].items() if v.numel() > 1 and v.shape[0] != nf},
+                                  strict=False)
+            for p in model.parameters():
                 p.requires_grad = True if p.shape[0] == nf else False
 
         else:  # resume from latest.pt