Remove image_encoder.py and update configuration_dpt.py

Removes the unused image_encoder.py file and uploads configuration_dpt.py.

configuration_dpt.py

@@ -22,7 +22,7 @@ class TIPSv2DPTConfig(PretrainedConfig):
         neck_hidden_sizes=(96, 192, 384, 768),
         fusion_hidden_size=256,
         reassemble_factors=(4, 2, 1, 0.5),
-        readout_act="gelu_pytorch_tanh",
+        readout_type="project",
         num_depth_bins=256,
         min_depth=1e-3,
         max_depth=10.0,
@@ -39,7 +39,7 @@ class TIPSv2DPTConfig(PretrainedConfig):
         self.channels = fusion_hidden_size
         self.post_process_channels = list(neck_hidden_sizes)
         self.block_indices = [out_index - 1 for out_index in out_indices]
-        self.readout_type = "project"
+        self.readout_type = readout_type
         self.num_depth_bins = num_depth_bins
         self.min_depth = min_depth
         self.max_depth = max_depth