Classify images with pre-trained ResNet in Pytorch

Image-Classification

from IPython.display import display # to display images
import torch
from torchvision import transforms
from PIL import Image
import torchvision.models as models
import requests
from io import BytesIO
import urllib.request, json
import numpy as np
normalize = transforms.Normalize(
   mean=[0.485, 0.456, 0.406],
   std=[0.229, 0.224, 0.225]
)
preprocess = transforms.Compose([
   transforms.Resize(256),
   transforms.CenterCrop(224),
   transforms.ToTensor(),
   normalize
])



def get_classification(url):

    response = requests.get(url)
    img = Image.open(BytesIO(response.content)).convert("RGB")
    #display(img)
    resnet_model = models.resnet152(pretrained=True, num_classes=1000)
    resnet_model.eval()
    img_tensor = preprocess(img).float()
    img_tensor = img_tensor.unsqueeze_(0)
    fc_out = resnet_model(img_tensor)
    output = fc_out.detach().numpy()
    return output


# downalod imagenet labels

with urllib.request.urlopen('https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json') as url:
   class_idx = json.loads(url.read().decode())

idx2label = [class_idx[str(k)][1] for k in range(len(class_idx))]


#images
img1_url= 'https://i.postimg.cc/SRRHTcgG/Screen-Shot-2019-01-25-at-11-56-52.png'
img2_url = 'https://i.postimg.cc/hG9nT8Yz/Screen-Shot-2019-01-25-at-12-09-15.png'

# predict classes
predition_img1 = get_classification(img1_url)
predition_img2 = get_classification(img2_url)


# print top10 predictions

predition_img1_sorted_top10 = np.argsort(predition_img1,axis=1)[0].tolist()[-10:]

print('Prediction for image %s'%(img1_url))
print('*'*20)
for idx in predition_img1_sorted_top10:
    print(idx2label[idx],predition_img1[0][idx])
print('*'*20)



predition_img2_sorted_top10 = np.argsort(predition_img2,axis=1)[0].tolist()[-10:]

print('Prediction for image %s'%(img2_url))
print('*'*20)
for idx in predition_img2_sorted_top10:
    print(idx2label[idx],predition_img2[0][idx])
print('*'*20)

      
#Prediction for image https://i.postimg.cc/SRRHTcgG/Screen-Shot-2019-01-25-at-11-56-52.png
# ********************
# otter 5.5231414
# beaver 5.572069
# bubble 5.6445265
# goose 5.6843433
# fountain 5.958186
# hippopotamus 6.0290885
# platypus 6.246428
# flamingo 6.2740183
# black_swan 7.0539513
# American_alligator 7.1412888
# ********************
#Prediction for image https://i.postimg.cc/hG9nT8Yz/Screen-Shot-2019-01-25-at-12-09-15.png
# ********************
# Weimaraner 8.421744
# American_Staffordshire_terrier 8.620165
# black-and-tan_coonhound 8.720882
# Mexican_hairless 9.13973
# miniature_pinscher 9.583361
# Rhodesian_ridgeback 9.651596
# Chihuahua 9.661165
# redbone 10.895191
# German_short-haired_pointer 11.339299
# bluetick 12.9905615
# ********************