Upload 3 files

app.py

@@ -0,0 +1,95 @@
+import cv2
+import os
+import random
+import numpy as np
+from matplotlib import pyplot as plt
+
+from tensorflow.keras.layers import Layer
+import tensorflow as tf
+
+import gradio as gr
+
+import uuid
+
+
+def preprocess(file_path):
+    # Read in image from file path
+    byte_img = tf.io.read_file(file_path)
+    # Load in the image
+    img = tf.io.decode_jpeg(byte_img)
+    # Preprocessing steps - resizing the image to be 100x100x3
+    img = tf.image.resize(img, (100, 100))
+    # Scale image to be between 0 and 1
+    img = img / 255.0
+    # Return image
+    return img
+
+# Siamese L1 Distance class
+class L1Dist(Layer):
+    # Init method - inheritance
+    def __init__(self, **kwargs):
+        super().__init__()
+    # Magic happens here - similarity calculation
+    def call(self, input_embedding, validation_embedding):
+        return tf.math.abs(input_embedding - validation_embedding)
+    
+# Reload model
+siamese_model = tf.keras.models.load_model('siamesemodelv2.h5',
+                                           custom_objects={'L1Dist': L1Dist, 'BinaryCrossentropy': tf.losses.BinaryCrossentropy})
+
+def verify(model, detection_threshold, verification_threshold):
+    # Build results array
+    results = []
+    ver_images= []
+    for file in os.listdir(os.path.join('application_data', 'verification_images')):
+      if file[len(file)-4:] == '.jpg':
+        ver_images.append(file)
+    ver_images_len= len(ver_images)
+    for image in ver_images:
+        input_img = preprocess(os.path.join('application_data', 'input_image', 'input_image.jpg'))
+        validation_img = preprocess(os.path.join('application_data', 'verification_images', image))
+        # Make Predictions 
+        result = model.predict(list(np.expand_dims([input_img, validation_img], axis=1)))
+        results.append(result)
+    
+    # Detection Threshold: Metric above which a prediciton is considered positive 
+    detection = np.sum(np.array(results) > detection_threshold)
+    print('Detection -- ',detection)
+    # Verification Threshold: Proportion of positive predictions / total positive samples 
+    verification = detection / len(ver_images) 
+    print('Verification -- ',verification)
+    verified = verification > verification_threshold
+    
+    return detection,verification,verified,ver_images_len
+
+def take_ver_images (image):
+    count =0
+    cv2.imwrite(os.path.join('application_data', 'verification_images', '{}.jpg'.format(uuid.uuid1())), image)
+    count+=1
+    return '{} Image Collected!'.format(count)
+def verify_identity (image):
+  cv2.imwrite(os.path.join('application_data', 'input_image', 'input_image.jpg'), image)
+  detection,verification,verified,ver_images_len = verify(siamese_model, 0.8, 0.7)
+  # return {'Detections': int(detection),'Confidence': float(verification),'Verified': bool(verified)}
+  return 'Detected {} out of {} samples'.format(detection,ver_images_len),float(verification),'Verified' if verified else 'Not Verified'
+
+with gr.Blocks() as demo:
+    with gr.Tab("Setup Identity"):
+      with gr.Row():
+        input_img = gr.inputs.Image(shape=(100,100))
+      output_text = gr.outputs.Label()
+      with gr.Row():
+        submit_btn = gr.Button("Add Image")
+        submit_btn.click(take_ver_images, inputs=input_img,outputs=output_text)
+    with gr.Tab("Identify"):
+      with gr.Row():
+        input_img1 = gr.inputs.Image(shape=(100,100))
+      output_text3 = gr.outputs.Label(label='Verification Result')
+      with gr.Row():
+        output_text1 = gr.outputs.Label(label='Detections')
+        output_text2 = gr.outputs.Label(label='Confidence')
+        
+      with gr.Row():
+        submit_btn1 = gr.Button("Verify")
+        submit_btn1.click(verify_identity, inputs=input_img1,outputs=[output_text1,output_text2,output_text3])
+demo.launch(debug=True)

requirements.txt

@@ -0,0 +1 @@
+gradio

siamesemodelv2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b167d0780c82879e76b0c3eebb42ae15d18b20bbca27108a3a6d249d478466e
+size 155886704