siompler sample script

Author: alvinwan

nbdt/bin/nbdt

@@ -3,16 +3,12 @@
 
 from nbdt.model import SoftNBDT
 from pytorchcv.models.wrn_cifar import wrn28_10_cifar10
-from PIL import Image
-from urllib.request import urlopen, Request
 from torchvision import transforms
-import io
+from nbdt.utils import DATASET_TO_CLASSES
 import sys
 
 assert len(sys.argv) > 1, "Need to pass image URL or image path as argument"
 
-classes = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']
-
 # load pretrained NBDT
 model = wrn28_10_cifar10()
 model = SoftNBDT(
@@ -22,27 +18,17 @@ model = SoftNBDT(
   pretrained=True,
   arch='wrn28_10_cifar10')
 
-# load image
-path = sys.argv[1]
-headers = {
-  'User-Agent': 'Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2228.0 Safari/537.3'
-}
-if 'http' in path:
-  request = Request(path, headers=headers)
-  file = io.BytesIO(urlopen(request).read())
-else:
-  file = path
-im = Image.open(file)
-
-# transform image
+# load + transform image
+im = load_image_from_path(sys.argv[1])
 transforms = transforms.Compose([
   transforms.Resize(32),
   transforms.CenterCrop(32),
-  transforms.ToTensor()
+  transforms.ToTensor(),
+  transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
 ])
 x = transforms(im)[None]
 
 # run inference
 outputs = model(x)
-cls = classes[outputs[0]]
+cls = DATASET_TO_CLASSES['CIFAR10'][outputs[0]]
 print(cls)

nbdt/utils.py

@@ -9,9 +9,12 @@
 import math
 import numpy as np
 
+from urllib.request import urlopen, Request
+from PIL import Image
 import torch.nn as nn
 import torch.nn.init as init
 from pathlib import Path
+import io
 
 # tree-generation consntants
 METHODS = ('wordnet', 'random', 'induced')
@@ -22,6 +25,12 @@
     'TinyImagenet200': 200,
     'Imagenet1000': 1000
 }
+DATASET_TO_CLASSES = {
+    'CIFAR10': [
+        'airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog',
+        'horse', 'ship', 'truck'
+    ]
+}
 
 
 def fwd():
@@ -61,6 +70,19 @@ def populate_kwargs(args, kwargs, object, name='Dataset', keys=(), globals={}):
                 f'{key}: {value}')
 
 
+def load_image_from_path(path):
+    """Path can be local or a URL"""
+    headers = {
+      'User-Agent': 'Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2228.0 Safari/537.3'
+    }
+    if 'http' in path:
+      request = Request(path, headers=headers)
+      file = io.BytesIO(urlopen(request).read())
+    else:
+      file = path
+    return Image.open(file)
+
+
 class Colors:
     RED = '\x1b[31m'
     GREEN = '\x1b[32m'