Adapt for Fast Dash

Dockerfile

@@ -56,8 +56,7 @@ RUN conda install -c conda-forge gradio -y
 
 WORKDIR /home/user
 
-RUN --mount=type=secret,id=git_token,mode=0444,required=true \
-    git clone --branch mmseg-only https://$(cat /run/secrets/git_token)@github.com/NASA-IMPACT/hls-foundation-os.git
+RUN git clone --branch mmseg-only https://github.com/NASA-IMPACT/hls-foundation-os.git
 
 
 WORKDIR hls-foundation-os 
@@ -66,7 +65,7 @@ RUN pip3 install -e .
 
 RUN mim install mmcv-full==1.6.2 -f https://download.openmmlab.com/mmcv/dist/11.5/1.11.0/index.html
 
-RUN pip3 install rasterio scikit-image
+RUN pip3 install rasterio scikit-image fast-dash gunicorn
 # Set the working directory to the user's home directory
 WORKDIR $HOME/app
 
@@ -75,4 +74,4 @@ ENV LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/code/miniconda/lib"
 # Copy the current directory contents into the container at $HOME/app setting the owner to the user
 COPY --chown=user . $HOME/app
 
-CMD ["python3", "app.py"]
+CMD exec gunicorn app:server --bind :7860

app.py

@@ -197,36 +197,38 @@ custom_test_pipeline=process_test_pipeline(model.cfg.data.test.pipeline, None)
 
 func = partial(inference_on_file, model=model, custom_test_pipeline=custom_test_pipeline)
 
-with gr.Blocks() as demo:
-   
-    gr.Markdown(value='# Prithvi sen1floods11')
-    gr.Markdown(value='''Prithvi is a first-of-its-kind temporal Vision transformer pretrained by the IBM and NASA team on continental US Harmonised Landsat Sentinel 2 (HLS) data. This demo showcases how the model was finetuned to detect water at a higher resolution than it was trained on (i.e. 10m versus 30m) using Sentinel 2 imagery from on the [sen1floods11 dataset](https://github.com/cloudtostreet/Sen1Floods11). More detailes can be found [here](https://huggingface.co/ibm-nasa-geospatial/Prithvi-100M-sen1floods11).\n
-    The user needs to provide a Sentinel 2 image with all the 12 bands (in the usual Sentinel 2) order in reflectance units multiplied by 10,000 (e.g. to save on space), with the code that is going to pull up Blue, Green, Red, Narrow NIR, SWIR, SWIR 2.
-    ''')
-    with gr.Row():
-        with gr.Column():
-            inp = gr.File()
-            btn = gr.Button("Submit")
-    
-    with gr.Row():
-        gr.Markdown(value='### Input RGB')
-        gr.Markdown(value='### Model prediction (Black: Land; White: Water)')
+
+if __name__ == "__main__":
+    with gr.Blocks() as demo:
+    
+        gr.Markdown(value='# Prithvi sen1floods11')
+        gr.Markdown(value='''Prithvi is a first-of-its-kind temporal Vision transformer pretrained by the IBM and NASA team on continental US Harmonised Landsat Sentinel 2 (HLS) data. This demo showcases how the model was finetuned to detect water at a higher resolution than it was trained on (i.e. 10m versus 30m) using Sentinel 2 imagery from on the [sen1floods11 dataset](https://github.com/cloudtostreet/Sen1Floods11). More detailes can be found [here](https://huggingface.co/ibm-nasa-geospatial/Prithvi-100M-sen1floods11).\n
+        The user needs to provide a Sentinel 2 image with all the 12 bands (in the usual Sentinel 2) order in reflectance units multiplied by 10,000 (e.g. to save on space), with the code that is going to pull up Blue, Green, Red, Narrow NIR, SWIR, SWIR 2.
+        ''')
+        with gr.Row():
+            with gr.Column():
+                inp = gr.File()
+                btn = gr.Button("Submit")
         
-    with gr.Row():
-        out1=gr.Image(image_mode='RGB')
-        out2 = gr.Image(image_mode='L')
-    
-    btn.click(fn=func, inputs=inp, outputs=[out1, out2])
-    
-    with gr.Row():
-        gr.Examples(examples=["India_900498_S2Hand.tif",
-                              "Spain_7370579_S2Hand.tif",
-                              "USA_430764_S2Hand.tif"],
-                    inputs=inp,
-                    outputs=[out1, out2],
-                    preprocess=preprocess_example,
-                    fn=func,
-                    cache_examples=True,
-    )
-
-demo.launch() 
+        with gr.Row():
+            gr.Markdown(value='### Input RGB')
+            gr.Markdown(value='### Model prediction (Black: Land; White: Water)')
+            
+        with gr.Row():
+            out1=gr.Image(image_mode='RGB')
+            out2 = gr.Image(image_mode='L')
+        
+        btn.click(fn=func, inputs=inp, outputs=[out1, out2])
+        
+        with gr.Row():
+            gr.Examples(examples=["India_900498_S2Hand.tif",
+                                "Spain_7370579_S2Hand.tif",
+                                "USA_430764_S2Hand.tif"],
+                        inputs=inp,
+                        outputs=[out1, out2],
+                        preprocess=preprocess_example,
+                        fn=func,
+                        cache_examples=True,
+        )
+
+    demo.launch(server_port=5001) 

fast_dash_app.py

@@ -0,0 +1,25 @@
+from fast_dash import FastDash, UploadImage, Image, Upload
+import PIL
+import numpy as np
+from app import func
+import tempfile
+import base64
+
+def generate_image(input_tiff_image: Upload) -> (Image, Image):
+
+    with tempfile.NamedTemporaryFile(delete=False) as f:
+        contents = input_tiff_image.encode("utf8").split(b";base64,")[1]
+        f.write(base64.decodebytes(contents))
+
+    input_image, output_image = func(f)
+
+    # Convert numpy arrays to PIL images
+    input_image = PIL.Image.fromarray(np.uint8(input_image)).convert('RGB')
+    output_image = PIL.Image.fromarray(np.uint8(output_image)).convert('RGB')
+
+    return input_image, output_image
+
+app = FastDash(generate_image, port=8000)
+
+if __name__ == "__main__":
+    app.run()

fastapp.py

@@ -0,0 +1,42 @@
+import os
+from fast_dash import FastDash, UploadImage, Image, Upload
+import PIL
+import numpy as np
+
+from app import func
+import tempfile
+import base64
+
+examples = ["India_900498_S2Hand", "Spain_7370579_S2Hand", "USA_430764_S2Hand"]
+
+def detect_water(select_an_example: str = examples, input_tiff_image: Upload = None) -> (Image, Image):
+    """
+    NASA and IBM recently uploaded their foundation model on Hugging Face, Pritivi, at https://huggingface.co/ibm-nasa-geospatial. 
+    This demo, built with Fast Dash, showcases a version of Prithvi that they finetuned to detect water from satellite images.
+    Select an example or upload your own TIFF image.
+    If uploading your own image, read the format details at https://huggingface.co/ibm-nasa-geospatial/Prithvi-100M-sen1floods11.
+    """
+
+    # If example is selected
+    if input_tiff_image is None:
+        input_satellite_image = PIL.Image.open(os.path.join("examples", f"{select_an_example}.png"))
+        water_prediction_mask = PIL.Image.open(os.path.join("examples", f"{select_an_example}_result.png"))
+
+    # If file is uploaded, run inference
+    else:
+        with tempfile.NamedTemporaryFile(delete=False) as f:
+            contents = input_tiff_image.encode("utf8").split(b";base64,")[1]
+            f.write(base64.decodebytes(contents))
+
+        input_satellite_image, water_prediction_mask = func(f)
+
+        input_satellite_image = PIL.Image.fromarray(np.uint8(input_satellite_image)).convert('RGB')
+        water_prediction_mask = PIL.Image.fromarray(np.uint8(water_prediction_mask)).convert('RGB')
+
+    return input_satellite_image, water_prediction_mask
+
+app = FastDash(detect_water, theme="cosmo", port=7860)
+server = app.server
+
+if __name__ == "__main__":
+    app.run()