app.py

import tensorflow as tf
import tensorflow_hub as hub

import requests
from PIL import Image
from io import BytesIO

import matplotlib.pyplot as plt
import numpy as np
import gradio as gr

#@title Helper functions for loading image (hidden)

original_image_cache = {}

def preprocess_image(image):
  image = np.array(image)
  # reshape into shape [batch_size, height, width, num_channels]
  img_reshaped = tf.reshape(image, [1, image.shape[0], image.shape[1], image.shape[2]])
  # Use `convert_image_dtype` to convert to floats in the [0,1] range.
  image = tf.image.convert_image_dtype(img_reshaped, tf.float32)
  return image

def load_image_from_url(img_url):
  """Returns an image with shape [1, height, width, num_channels]."""
  user_agent = {'User-agent': 'Colab Sample (https://tensorflow.org)'}
  response = requests.get(img_url, headers=user_agent)
  image = Image.open(BytesIO(response.content))
  image = preprocess_image(image)
  return image

def load_image(image_url, image_size=256, dynamic_size=False, max_dynamic_size=512):
  """Loads and preprocesses images."""
  # Cache image file locally.
  if image_url in original_image_cache:
    img = original_image_cache[image_url]
  elif image_url.startswith('https://'):
    img = load_image_from_url(image_url)
  else:
    fd = tf.io.gfile.GFile(image_url, 'rb')
    img = preprocess_image(Image.open(fd))
  original_image_cache[image_url] = img
  # Load and convert to float32 numpy array, add batch dimension, and normalize to range [0, 1].
  img_raw = img
  if tf.reduce_max(img) > 1.0:
    img = img / 255.
  if len(img.shape) == 3:
    img = tf.stack([img, img, img], axis=-1)
  if not dynamic_size:
    img = tf.image.resize_with_pad(img, image_size, image_size)
  elif img.shape[1] > max_dynamic_size or img.shape[2] > max_dynamic_size:
    img = tf.image.resize_with_pad(img, max_dynamic_size, max_dynamic_size)
  return img, img_raw



image_size = 224
dynamic_size = False

model_name = "efficientnetv2-b0" 

model_handle_map = {
  "efficientnetv2-s": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_s/classification/2",
  "efficientnetv2-m": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_m/classification/2",
  "efficientnetv2-l": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_l/classification/2",
  "efficientnetv2-s-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_s/classification/2",
  "efficientnetv2-m-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_m/classification/2",
  "efficientnetv2-l-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_l/classification/2",
  "efficientnetv2-xl-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_xl/classification/2",
  "efficientnetv2-b0-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_b0/classification/2",
  "efficientnetv2-b1-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_b1/classification/2",
  "efficientnetv2-b2-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_b2/classification/2",
  "efficientnetv2-b3-21k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_b3/classification/2",
  "efficientnetv2-s-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_s/classification/2",
  "efficientnetv2-m-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_m/classification/2",
  "efficientnetv2-l-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_l/classification/2",
  "efficientnetv2-xl-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_xl/classification/2",
  "efficientnetv2-b0-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_b0/classification/2",
  "efficientnetv2-b1-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_b1/classification/2",
  "efficientnetv2-b2-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_b2/classification/2",
  "efficientnetv2-b3-21k-ft1k": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet21k_ft1k_b3/classification/2",
  "efficientnetv2-b0": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_b0/classification/2",
  "efficientnetv2-b1": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_b1/classification/2",
  "efficientnetv2-b2": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_b2/classification/2",
  "efficientnetv2-b3": "https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_b3/classification/2",
  "efficientnet_b0": "https://tfhub.dev/tensorflow/efficientnet/b0/classification/1",
  "efficientnet_b1": "https://tfhub.dev/tensorflow/efficientnet/b1/classification/1",
  "efficientnet_b2": "https://tfhub.dev/tensorflow/efficientnet/b2/classification/1",
  "efficientnet_b3": "https://tfhub.dev/tensorflow/efficientnet/b3/classification/1",
  "efficientnet_b4": "https://tfhub.dev/tensorflow/efficientnet/b4/classification/1",
  "efficientnet_b5": "https://tfhub.dev/tensorflow/efficientnet/b5/classification/1",
  "efficientnet_b6": "https://tfhub.dev/tensorflow/efficientnet/b6/classification/1",
  "efficientnet_b7": "https://tfhub.dev/tensorflow/efficientnet/b7/classification/1",
  "bit_s-r50x1": "https://tfhub.dev/google/bit/s-r50x1/ilsvrc2012_classification/1",
  "inception_v3": "https://tfhub.dev/google/imagenet/inception_v3/classification/4",
  "inception_resnet_v2": "https://tfhub.dev/google/imagenet/inception_resnet_v2/classification/4",
  "resnet_v1_50": "https://tfhub.dev/google/imagenet/resnet_v1_50/classification/4",
  "resnet_v1_101": "https://tfhub.dev/google/imagenet/resnet_v1_101/classification/4",
  "resnet_v1_152": "https://tfhub.dev/google/imagenet/resnet_v1_152/classification/4",
  "resnet_v2_50": "https://tfhub.dev/google/imagenet/resnet_v2_50/classification/4",
  "resnet_v2_101": "https://tfhub.dev/google/imagenet/resnet_v2_101/classification/4",
  "resnet_v2_152": "https://tfhub.dev/google/imagenet/resnet_v2_152/classification/4",
  "nasnet_large": "https://tfhub.dev/google/imagenet/nasnet_large/classification/4",
  "nasnet_mobile": "https://tfhub.dev/google/imagenet/nasnet_mobile/classification/4",
  "pnasnet_large": "https://tfhub.dev/google/imagenet/pnasnet_large/classification/4",
  "mobilenet_v2_100_224": "https://tfhub.dev/google/imagenet/mobilenet_v2_100_224/classification/4",
  "mobilenet_v2_130_224": "https://tfhub.dev/google/imagenet/mobilenet_v2_130_224/classification/4",
  "mobilenet_v2_140_224": "https://tfhub.dev/google/imagenet/mobilenet_v2_140_224/classification/4",
  "mobilenet_v3_small_100_224": "https://tfhub.dev/google/imagenet/mobilenet_v3_small_100_224/classification/5",
  "mobilenet_v3_small_075_224": "https://tfhub.dev/google/imagenet/mobilenet_v3_small_075_224/classification/5",
  "mobilenet_v3_large_100_224": "https://tfhub.dev/google/imagenet/mobilenet_v3_large_100_224/classification/5",
  "mobilenet_v3_large_075_224": "https://tfhub.dev/google/imagenet/mobilenet_v3_large_075_224/classification/5",
}

model_image_size_map = {
  "efficientnetv2-s": 384,
  "efficientnetv2-m": 480,
  "efficientnetv2-l": 480,
  "efficientnetv2-b0": 224,
  "efficientnetv2-b1": 240,
  "efficientnetv2-b2": 260,
  "efficientnetv2-b3": 300,
  "efficientnetv2-s-21k": 384,
  "efficientnetv2-m-21k": 480,
  "efficientnetv2-l-21k": 480,
  "efficientnetv2-xl-21k": 512,
  "efficientnetv2-b0-21k": 224,
  "efficientnetv2-b1-21k": 240,
  "efficientnetv2-b2-21k": 260,
  "efficientnetv2-b3-21k": 300,
  "efficientnetv2-s-21k-ft1k": 384,
  "efficientnetv2-m-21k-ft1k": 480,
  "efficientnetv2-l-21k-ft1k": 480,
  "efficientnetv2-xl-21k-ft1k": 512,
  "efficientnetv2-b0-21k-ft1k": 224,
  "efficientnetv2-b1-21k-ft1k": 240,
  "efficientnetv2-b2-21k-ft1k": 260,
  "efficientnetv2-b3-21k-ft1k": 300, 
  "efficientnet_b0": 224,
  "efficientnet_b1": 240,
  "efficientnet_b2": 260,
  "efficientnet_b3": 300,
  "efficientnet_b4": 380,
  "efficientnet_b5": 456,
  "efficientnet_b6": 528,
  "efficientnet_b7": 600,
  "inception_v3": 299,
  "inception_resnet_v2": 299,
  "mobilenet_v2_100_224": 224,
  "mobilenet_v2_130_224": 224,
  "mobilenet_v2_140_224": 224,
  "nasnet_large": 331,
  "nasnet_mobile": 224,
  "pnasnet_large": 331,
  "resnet_v1_50": 224,
  "resnet_v1_101": 224,
  "resnet_v1_152": 224,
  "resnet_v2_50": 224,
  "resnet_v2_101": 224,
  "resnet_v2_152": 224,
  "mobilenet_v3_small_100_224": 224,
  "mobilenet_v3_small_075_224": 224,
  "mobilenet_v3_large_100_224": 224,
  "mobilenet_v3_large_075_224": 224,
}

model_handle = model_handle_map[model_name]


max_dynamic_size = 512
if model_name in model_image_size_map:
  image_size = model_image_size_map[model_name]
  dynamic_size = False
  print(f"Images will be converted to {image_size}x{image_size}")
else:
  dynamic_size = True
  print(f"Images will be capped to a max size of {max_dynamic_size}x{max_dynamic_size}")

labels_file = "https://storage.googleapis.com/download.tensorflow.org/data/ImageNetLabels.txt"

#download labels and creates a maps
downloaded_file = tf.keras.utils.get_file("labels.txt", origin=labels_file)

classes = []

with open(downloaded_file) as f:
  labels = f.readlines()
  classes = [l.strip() for l in labels]


classifier = hub.load(model_handle)


def inference(img):
  image, original_image = load_image(img, image_size, dynamic_size, max_dynamic_size)  
  
  
  input_shape = image.shape
  warmup_input = tf.random.uniform(input_shape, 0, 1.0)
  warmup_logits = classifier(warmup_input).numpy()
  
  # Run model on image
  probabilities = tf.nn.softmax(classifier(image)).numpy()
  
  top_5 = tf.argsort(probabilities, axis=-1, direction="DESCENDING")[0][:5].numpy()
  np_classes = np.array(classes)
  
  # Some models include an additional 'background' class in the predictions, so
  # we must account for this when reading the class labels.
  includes_background_class = probabilities.shape[1] == 1001
  result = {}
  for i, item in enumerate(top_5):
    class_index = item if includes_background_class else item + 1
    line = f'({i+1}) {class_index:4} - {classes[class_index]}: {probabilities[0][top_5][i]}'
    result[classes[class_index]] = probabilities[0][top_5][i].item()
  return result

title="Image Classification"
description="Gradio Demo on Imagenet (ILSVRC-2012-CLS) classification with EfficientNet V2. To use it, simply upload your image or click on one of the examples to load them."
article = "<p style='text-align: center'><a href='https://tfhub.dev/google/imagenet/efficientnet_v2_imagenet1k_b0/classification/2' target='_blank'>Tensorflow Hub</a></p>"
examples=[['Elephant.jpeg']]
gr.Interface(inference,gr.inputs.Image(type="filepath"),"label",title=title,description=description,article=article,examples=examples).launch(enable_queue=True)