Add gradio demo

broadlume · Mar 10, 2023 · edb6daf · edb6daf
1 parent 2361c75
commit edb6daf
Show file tree

Hide file tree

Showing 7 changed files with 443 additions and 0 deletions.
diff --git a/README.md b/README.md
@@ -1,5 +1,6 @@
 # **ZoeDepth: Combining relative and metric depth** (Official implementation)  <!-- omit in toc -->
 [![Open In Collab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/isl-org/ZoeDepth)
+[![Open in Spaces](https://huggingface.co/datasets/huggingface/badges/raw/main/open-in-hf-spaces-sm.svg)](https://huggingface.co/spaces/shariqfarooq/ZoeDepth)
 
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT) ![PyTorch](https://img.shields.io/badge/PyTorch_v1.10.1-EE4C2C?&logo=pytorch&logoColor=white) 
 [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/zoedepth-zero-shot-transfer-by-combining/monocular-depth-estimation-on-nyu-depth-v2)](https://paperswithcode.com/sota/monocular-depth-estimation-on-nyu-depth-v2?p=zoedepth-zero-shot-transfer-by-combining)
@@ -25,6 +26,7 @@
   - [Evaluating offical models](#evaluating-offical-models)
   - [Evaluating local checkpoint](#evaluating-local-checkpoint)
 - [**Training**](#training)
+- [**Gradio demo**](#gradio-demo)
 - [**Citation**](#citation)
 
 
@@ -199,7 +201,17 @@ For training the Zoe-NK model:
 ```bash
 python train_mix.py -m zoedepth_nk --pretrained_resource=""
 ```
+## **Gradio demo**
+We provide a UI demo built using [gradio](https://gradio.app/). To get started, install UI requirements:
+```bash
+pip install -r ui/ui_requirements.txt
+```
+Then launch the gradio UI:
+```bash
+python -m ui.app
+```
 
+The UI is also hosted on HuggingFace🤗 [here](https://huggingface.co/spaces/shariqfarooq/ZoeDepth)
 ## **Citation**
 ```
 @misc{https://doi.org/10.48550/arxiv.2302.12288,

diff --git a/ui/app.py b/ui/app.py
@@ -0,0 +1,66 @@
+# MIT License
+
+# Copyright (c) 2022 Intelligent Systems Lab Org
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# File author: Shariq Farooq Bhat
+
+import gradio as gr
+import torch
+
+from .gradio_depth_pred import create_demo as create_depth_pred_demo
+from .gradio_im_to_3d import create_demo as create_im_to_3d_demo
+from .gradio_pano_to_3d import create_demo as create_pano_to_3d_demo
+
+
+css = """
+#img-display-container {
+    max-height: 50vh;
+    }
+#img-display-input {
+    max-height: 40vh;
+    }
+#img-display-output {
+    max-height: 40vh;
+    }
+    
+"""
+DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
+model = torch.hub.load('isl-org/ZoeDepth', "ZoeD_N", pretrained=True).to(DEVICE).eval()
+
+title = "# ZoeDepth"
+description = """Official demo for **ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth**.
+
+ZoeDepth is a deep learning model for metric depth estimation from a single image.
+
+Please refer to our [paper](https://arxiv.org/abs/2302.12288) or [github](https://github.com/isl-org/ZoeDepth) for more details."""
+
+with gr.Blocks(css=css) as demo:
+    gr.Markdown(title)
+    gr.Markdown(description)
+    with gr.Tab("Depth Prediction"):
+        create_depth_pred_demo(model)
+    with gr.Tab("Image to 3D"):
+        create_im_to_3d_demo(model)
+    with gr.Tab("360 Panorama to 3D"):
+        create_pano_to_3d_demo(model)
+
+if __name__ == '__main__':
+    demo.queue().launch()
diff --git a/ui/gradio_depth_pred.py b/ui/gradio_depth_pred.py
@@ -0,0 +1,52 @@
+# MIT License
+
+# Copyright (c) 2022 Intelligent Systems Lab Org
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# File author: Shariq Farooq Bhat
+
+import gradio as gr
+from zoedepth.utils.misc import colorize
+from PIL import Image
+import tempfile
+
+def predict_depth(model, image):
+    depth = model.infer_pil(image)
+    return depth
+
+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")
+        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")
+
+    def on_submit(image):
+        depth = predict_depth(model, image)
+        colored_depth = colorize(depth, cmap='gray_r')
+        tmp = tempfile.NamedTemporaryFile(suffix='.png', delete=False)
+        raw_depth = Image.fromarray((depth*256).astype('uint16'))
+        raw_depth.save(tmp.name)
+        return [colored_depth, tmp.name]
+
+    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])
diff --git a/ui/gradio_im_to_3d.py b/ui/gradio_im_to_3d.py
@@ -0,0 +1,93 @@
+# MIT License
+
+# Copyright (c) 2022 Intelligent Systems Lab Org
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# File author: Shariq Farooq Bhat
+
+import gradio as gr
+import numpy as np
+import trimesh
+from zoedepth.utils.geometry import depth_to_points, create_triangles
+from functools import partial
+import tempfile
+
+
+def depth_edges_mask(depth):
+    """Returns a mask of edges in the depth map.
+    Args:
+    depth: 2D numpy array of shape (H, W) with dtype float32.
+    Returns:
+    mask: 2D numpy array of shape (H, W) with dtype bool.
+    """
+    # Compute the x and y gradients of the depth map.
+    depth_dx, depth_dy = np.gradient(depth)
+    # Compute the gradient magnitude.
+    depth_grad = np.sqrt(depth_dx ** 2 + depth_dy ** 2)
+    # Compute the edge mask.
+    mask = depth_grad > 0.05
+    return mask
+
+
+def predict_depth(model, image):
+    depth = model.infer_pil(image)
+    return depth
+
+def get_mesh(model, image, keep_edges=False):
+    image.thumbnail((1024,1024))  # limit the size of the input image
+    depth = predict_depth(model, image)
+    pts3d = depth_to_points(depth[None])
+    pts3d = pts3d.reshape(-1, 3)
+
+    # Create a trimesh mesh from the points
+    # Each pixel is connected to its 4 neighbors
+    # colors are the RGB values of the image
+
+    verts = pts3d.reshape(-1, 3)
+    image = np.array(image)
+    if keep_edges:
+        triangles = create_triangles(image.shape[0], image.shape[1])
+    else:
+        triangles = create_triangles(image.shape[0], image.shape[1], mask=~depth_edges_mask(depth))
+    colors = image.reshape(-1, 3)
+    mesh = trimesh.Trimesh(vertices=verts, faces=triangles, vertex_colors=colors)
+
+    # Save as glb
+    glb_file = tempfile.NamedTemporaryFile(suffix='.glb', delete=False)
+    glb_path = glb_file.name
+    mesh.export(glb_path)
+    return glb_path
+
+def create_demo(model):
+
+    gr.Markdown("### Image to 3D mesh")
+    gr.Markdown("Convert a single 2D image to a 3D mesh")
+
+    with gr.Row():
+        image = gr.Image(label="Input Image", type='pil')
+        result = gr.Model3D(label="3d mesh reconstruction", clear_color=[
+                                                 1.0, 1.0, 1.0, 1.0])
+
+    checkbox = gr.Checkbox(label="Keep occlusion edges", value=False)
+    submit = gr.Button("Submit")
+    submit.click(partial(get_mesh, model), inputs=[image, checkbox], outputs=[result])
+    # examples = gr.Examples(examples=["examples/aerial_beach.jpeg", "examples/mountains.jpeg", "examples/person_1.jpeg", "examples/ancient-carved.jpeg"],
+    #                         inputs=[image])
+
diff --git a/ui/gradio_pano_to_3d.py b/ui/gradio_pano_to_3d.py
@@ -0,0 +1,120 @@
+# MIT License
+
+# Copyright (c) 2022 Intelligent Systems Lab Org
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# File author: Shariq Farooq Bhat
+
+import gradio as gr
+import numpy as np
+import trimesh
+from zoedepth.utils.geometry import create_triangles
+from functools import partial
+import tempfile
+
+def depth_edges_mask(depth):
+    """Returns a mask of edges in the depth map.
+    Args:
+    depth: 2D numpy array of shape (H, W) with dtype float32.
+    Returns:
+    mask: 2D numpy array of shape (H, W) with dtype bool.
+    """
+    # Compute the x and y gradients of the depth map.
+    depth_dx, depth_dy = np.gradient(depth)
+    # Compute the gradient magnitude.
+    depth_grad = np.sqrt(depth_dx ** 2 + depth_dy ** 2)
+    # Compute the edge mask.
+    mask = depth_grad > 0.05
+    return mask
+
+
+def pano_depth_to_world_points(depth):
+    """
+    360 depth to world points
+    given 2D depth is an equirectangular projection of a spherical image
+    Treat depth as radius
+
+    longitude : -pi to pi
+    latitude : -pi/2 to pi/2
+    """
+
+    # Convert depth to radius
+    radius = depth.flatten()
+
+    lon = np.linspace(-np.pi, np.pi, depth.shape[1])
+    lat = np.linspace(-np.pi/2, np.pi/2, depth.shape[0])
+
+    lon, lat = np.meshgrid(lon, lat)
+    lon = lon.flatten()
+    lat = lat.flatten()
+
+    # Convert to cartesian coordinates
+    x = radius * np.cos(lat) * np.cos(lon)
+    y = radius * np.cos(lat) * np.sin(lon)
+    z = radius * np.sin(lat)
+
+    pts3d = np.stack([x, y, z], axis=1)
+
+    return pts3d
+
+
+def predict_depth(model, image):
+    depth = model.infer_pil(image)
+    return depth
+
+def get_mesh(model, image, keep_edges=False):
+    image.thumbnail((1024,1024))  # limit the size of the image
+    depth = predict_depth(model, image)
+    pts3d = pano_depth_to_world_points(depth)
+
+    # Create a trimesh mesh from the points
+    # Each pixel is connected to its 4 neighbors
+    # colors are the RGB values of the image
+
+    verts = pts3d.reshape(-1, 3)
+    image = np.array(image)
+    if keep_edges:
+        triangles = create_triangles(image.shape[0], image.shape[1])
+    else:
+        triangles = create_triangles(image.shape[0], image.shape[1], mask=~depth_edges_mask(depth))
+    colors = image.reshape(-1, 3)
+    mesh = trimesh.Trimesh(vertices=verts, faces=triangles, vertex_colors=colors)
+
+    # Save as glb
+    glb_file = tempfile.NamedTemporaryFile(suffix='.glb', delete=False)
+    glb_path = glb_file.name
+    mesh.export(glb_path)
+    return glb_path
+
+def create_demo(model):
+    gr.Markdown("### Panorama to 3D mesh")
+    gr.Markdown("Convert a 360 spherical panorama to a 3D mesh")
+    gr.Markdown("ZoeDepth was not trained on panoramic images. It doesn't know anything about panoramas or spherical projection. Here, we just treat the estimated depth as radius and some projection errors are expected. Nonetheless, ZoeDepth still works surprisingly well on 360 reconstruction.")
+
+    with gr.Row():
+        input_image = gr.Image(label="Input Image", type='pil')
+        result = gr.Model3D(label="3d mesh reconstruction", clear_color=[
+                                                 1.0, 1.0, 1.0, 1.0])
+
+    checkbox = gr.Checkbox(label="Keep occlusion edges", value=True)
+    submit = gr.Button("Submit")
+    submit.click(partial(get_mesh, model), inputs=[input_image, checkbox], outputs=[result])
+    # examples = gr.Examples(examples=["examples/pano_1.jpeg", "examples/pano_2.jpeg", "examples/pano_3.jpeg"],
+    #                         inputs=[input_image])
diff --git a/ui/ui_requirements.txt b/ui/ui_requirements.txt
@@ -0,0 +1,2 @@
+gradio
+trimesh==3.9.42