app.py

import gradio as gr 
import tensorflow as tf
from keras.models import load_model
import os
os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"
os.environ["SM_FRAMEWORK"] = "tf.keras"
import segmentation_models as sm
from keras.metrics import MeanIoU

def jaccard_coef(y_true, y_pred):
    """
    Defines custom jaccard coefficient metric
    """
    
    y_true_flatten = K.flatten(y_true)
    y_pred_flatten = K.flatten(y_pred)
    intersection = K.sum(y_true_flatten * y_pred_flatten)
    final_coef_value = (intersection + 1.0) / (K.sum(y_true_flatten) + K.sum(y_pred_flatten) - intersection + 1.0)
    return final_coef_value

def real_dice_coeff(y_true, y_pred):
    smooth = 0.0001
    y_true_flatten = K.flatten(y_true)
    y_pred_flatten = K.flatten(y_pred)
    intersection = K.sum(y_true_flatten * y_pred_flatten)
    dice_score = (2.0 * intersection + smooth) / (K.sum(y_true_flatten) + K.sum(y_pred_flatten) + smooth)
    return dice_score


dice_loss = sm.losses.DiceLoss(class_weights = weights)
focal_loss = sm.losses.CategoricalFocalLoss()
TOTAL_LOSS_FACTOR = 5
total_loss = dice_loss + (TOTAL_LOSS_FACTOR * focal_loss)

metrics = [real_dice_coeff, tf.keras.metrics.MeanIoU(num_classes=2, sparse_y_true= False, sparse_y_pred=False, name="Mean IOU"), "accuracy", jaccard_coef, sm.metrics.FScore(threshold=0.6, name="Dice Coeficient")]


model = load_model('../../../fast-disk/w210-capstone/models/' + model_name, custom_objects={'dice_loss_plus_5focal_loss': total_loss, 'jaccard_coef': jaccard_coef, 'IOU score' : sm.metrics.IOUScore(threshold=0.9, name="IOU score"), 'Dice Coeficient' : sm.metrics.FScore(threshold=0.6, name="Dice Coeficient")}, compile=False)
model.compile(metrics=metrics)


# def greet(name):
#     return "Hello " + name + "!!"

# iface = gr.Interface(fn=greet, inputs="text", outputs="text")
# iface.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()