only cpu is used in gui mode fix and hacky gardio/timm fixes

README.md

@@ -42,15 +42,36 @@ torch.hub.help("intel-isl/MiDaS", "DPT_BEiT_L_384", force_reload=True)  # Trigge
 ```python
 import torch
 
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+
 repo = "isl-org/ZoeDepth"
 # Zoe_N
-model_zoe_n = torch.hub.load(repo, "ZoeD_N", pretrained=True)
+model_zoe_n = torch.hub.load(repo, "ZoeD_N", pretrained=False)
+pretrained_dict = torch.hub.load_state_dict_from_url('https://github.com/isl-org/ZoeDepth/releases/download/v1.0/ZoeD_M12_N.pt', map_location=DEVICE)
+model_zoe_n.load_state_dict(pretrained_dict['model'], strict=False)
+for b in model_zoe_n.core.core.pretrained.model.blocks:
+    b.drop_path = torch.nn.Identity()
+
+zoe = model_zoe_n.to(DEVICE)
 
 # Zoe_K
-model_zoe_k = torch.hub.load(repo, "ZoeD_K", pretrained=True)
+model_zoe_k = torch.hub.load(repo, "ZoeD_K", pretrained=False)
+pretrained_dict = torch.hub.load_state_dict_from_url('https://github.com/isl-org/ZoeDepth/releases/#download/v1.0/ZoeD_M12_K.pt', map_location=DEVICE)
+model_zoe_k.load_state_dict(pretrained_dict['model'], strict=False)
+for b in model_zoe_k.core.core.pretrained.model.blocks:
+    b.drop_path = torch.nn.Identity()
+
+zoe = model_zoe_k.to(DEVICE)
 
 # Zoe_NK
-model_zoe_nk = torch.hub.load(repo, "ZoeD_NK", pretrained=True)
+model_zoe_nk = torch.hub.load(repo, "ZoeD_NK", pretrained=False)
+pretrained_dict = torch.hub.load_state_dict_from_url('https://github.com/isl-org/ZoeDepth/releases/download/v1.0/ZoeD_M12_NK.pt', map_location=DEVICE)
+model_zoe_nk.load_state_dict(pretrained_dict['model'], strict=False)
+for b in model_zoe_nk.core.core.pretrained.model.blocks:
+    b.drop_path = torch.nn.Identity()
+
+zoe = model_zoe_nk.to(DEVICE)
+
 ```
 ### Using local copy
 Clone this repo:
@@ -87,8 +108,8 @@ model_zoe_nk = build_model(conf)
 ### Using ZoeD models to predict depth 
 ```python
 ##### sample prediction
-DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
-zoe = model_zoe_n.to(DEVICE)
+
+
 
 
 # Local file

ui/app.py

@@ -43,7 +43,18 @@
 
 """
 DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
-model = torch.hub.load('isl-org/ZoeDepth', "ZoeD_N", pretrained=True).to(DEVICE).eval()
+
+# https://github.com/isl-org/ZoeDepth/issues/82#issuecomment-1779799540
+repo = "isl-org/ZoeDepth"
+
+model_zoe_n = torch.hub.load(repo, "ZoeD_N", pretrained=False)
+pretrained_dict = torch.hub.load_state_dict_from_url('https://github.com/isl-org/ZoeDepth/releases/download/v1.0/ZoeD_M12_N.pt', map_location=DEVICE)
+model_zoe_n.load_state_dict(pretrained_dict['model'], strict=False)
+for b in model_zoe_n.core.core.pretrained.model.blocks:
+    b.drop_path = torch.nn.Identity()
+
+model = model_zoe_n
+zoe = model_zoe_n.to(DEVICE)
 
 title = "# ZoeDepth"
 description = """Official demo for **ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth**.
@@ -63,4 +74,4 @@
         create_pano_to_3d_demo(model)
 
 if __name__ == '__main__':
-    demo.queue().launch()
+    demo.queue().launch()

ui/gradio_depth_pred.py

@@ -34,7 +34,7 @@ def predict_depth(model, image):
 def create_demo(model):
     gr.Markdown("### Depth Prediction demo")
     with gr.Row():
-        input_image = gr.Image(label="Input Image", type='pil', elem_id='img-display-input').style(height="auto")
+        input_image = gr.Image(label="Input Image", type='pil', elem_id='img-display-input')
         depth_image = gr.Image(label="Depth Map", elem_id='img-display-output')
     raw_file = gr.File(label="16-bit raw depth, multiplier:256")
     submit = gr.Button("Submit")
@@ -49,4 +49,4 @@ def on_submit(image):
 
     submit.click(on_submit, inputs=[input_image], outputs=[depth_image, raw_file])
     # examples = gr.Examples(examples=["examples/person_1.jpeg", "examples/person_2.jpeg", "examples/person-leaves.png", "examples/living-room.jpeg"],
-    #                        inputs=[input_image])
+    #                        inputs=[input_image])

zoedepth/models/base_models/midas.py

@@ -170,7 +170,7 @@ def get_size(self, width, height):
 
     def __call__(self, x):
         width, height = self.get_size(*x.shape[-2:][::-1])
-        return nn.functional.interpolate(x, (height, width), mode='bilinear', align_corners=True)
+        return nn.functional.interpolate(x, (int(height), int(width)), mode='bilinear', align_corners=True)
 
 class PrepForMidas(object):
     def __init__(self, resize_mode="minimal", keep_aspect_ratio=True, img_size=384, do_resize=True):