Fix Resnet pretrained weights with in_chans argument using timm adapt…

…_input_conv (#743)

flash/image/classification/backbones/resnet.py

@@ -25,8 +25,12 @@
 from torch.hub import load_state_dict_from_url
 
 from flash.core.registry import FlashRegistry
+from flash.core.utilities.imports import _TIMM_AVAILABLE
 from flash.core.utilities.url_error import catch_url_error
 
+if _TIMM_AVAILABLE:
+    from timm.models.helpers import adapt_input_conv
+
 
 def conv3x3(in_planes: int, out_planes: int, stride: int = 1, groups: int = 1, dilation: int = 1) -> nn.Conv2d:
     """3x3 convolution with padding."""
@@ -351,6 +355,8 @@ def _resnet(
             )
 
     if model_weights is not None:
+        in_chans = backbone.conv1.weight.shape[1]
+        model_weights["conv1.weight"] = adapt_input_conv(in_chans, model_weights["conv1.weight"])
         backbone.load_state_dict(model_weights)
 
     return backbone, num_features