attempt tuples

app.py

@@ -46,9 +46,6 @@ model = load_model('unet_contrails_model.keras', custom_objects={'dice_loss_plus
 model.compile(metrics=metrics)
 
 
-
-
-
 # def greet(name):
 #     return "Hello " + name + "!!"
 
@@ -58,7 +55,7 @@ model.compile(metrics=metrics)
 def predict(ash_image, model=model):
     #label = np.load(label_image)
     # ash_image = np.load('images/552609781892851211.npy')
-    ash_image = ash_image
+    ash_image = np.load(ash_image)
     y_pred = model.predict(ash_image.reshape(1,256, 256, 3))
     prediction = np.argmax(y_pred[0], axis=2).reshape(256,256)
     #intersection = label & prediction
@@ -69,8 +66,6 @@ def predict(ash_image, model=model):
     return(ash_image, seg_info)
 
 
-
-
 if __name__ == "__main__":
     class2hexcolor = {"contrails": "#FF0000"}
 
@@ -78,41 +73,16 @@ if __name__ == "__main__":
         gr.Markdown("""<h1><center>Predict Contrails in Satellite Images</center></h1>""")
         with gr.Row():
             img_input = gr.Image(type="numpy", height=256, width=256, label="Input image")
+            file_input = gr.File()
             img_output = gr.AnnotatedImage(label="Predictions", height=256, width=256, color_map=class2hexcolor)
 
         section_btn = gr.Button("Generate Predictions")
-        section_btn.click(partial(predict, model=model), img_input, img_output)
+        section_btn.click(partial(predict, model=model), img_input[1], img_output)
 
         images_dir = glob(os.path.join(os.getcwd(), "images") + os.sep + "*.png")
-        examples = [i for i in images_dir]
-        gr.Examples(examples=examples, inputs=img_input, outputs=img_output)
+        files_dir = glob(os.path.join(os.getcwd(), "numpy_files") + os.sep + "*.npy")
+       # x = (images_dir,files_dir)
+        examples = [[i for i in images_dir],[f for f in files_dir]]
+        gr.Examples(examples=examples, inputs=[img_input,file_input], outputs=img_output)
 
     demo.launch()
-
-
-
-# first we need to load the model from somewhere - probably by using model.load on a keras file (which could be saved in our huggingface space repo)
-# once we have a model, we use tensorflow to run inference and create a visualization of the output mask
-# just like we did in our main code like such:
-# model_name = "something.keras" (which should be saved in our HF repo)
-# model = load_model(model_name)
-# model.compile(?)
-
-
-# then we save that inference + viz task in a function called "make mask" which takes an input image, processes it, runs the model on it, and returns a mask image 
-# within this function we can use functions like display_image_and_label (and the functions that that function calls to make predictions) 
-# instead of just printing the visualization, we have the function return the image output
-
-
-#description = "<p style='text-align: center'>Gradio demo for identifying contrails in satellite images. </p>"
-#css = "#0 {object-fit: contain;} #1 {object-fit: contain;}"
-#demo = gr.Interface(fn=make_mask, 
- #                   title=title, 
-  #                  description=description,
-   #                 inputs= # either the use input or one we save as a file, 
-    #                outputs=gr.Image(elem_id=1, show_label=False),
-     #               css=css, 
-      #              examples=examples, 
-       #             cache_examples=True,
-        #            allow_flagging='never')
-#demo.launch()