Skip to content

Latest commit

 

History

History
113 lines (96 loc) · 3.9 KB

pytorch_vision_ibnnet.md

File metadata and controls

113 lines (96 loc) · 3.9 KB
Raw
layout background-class body-class title summary category image author tags github-link github-id featured_image_1 featured_image_2 accelerator order
hub_detail
hub-background
hub
IBN-Net
Networks with domain/appearance invariance
researchers
ibnnet.png
Xingang Pan
vision
XingangPan/IBN-Net
ibnnet.png
no-image
cuda-optional
10
 
import torch
model = torch.hub.load('XingangPan/IBN-Net', 'resnet50_ibn_a', pretrained=True)
model.eval()

All pre-trained models expect input images normalized in the same way, i.e. mini-batches of 3-channel RGB images of shape (3 x H x W), where H and W are expected to be at least 224. The images have to be loaded in to a range of [0, 1] and then normalized using mean = [0.485, 0.456, 0.406] and std = [0.229, 0.224, 0.225].

Here's a sample execution.

# Download an example image from the pytorch website
import urllib
url, filename = ("https://github.com/pytorch/hub/raw/master/images/dog.jpg", "dog.jpg")
try: urllib.URLopener().retrieve(url, filename)
except: urllib.request.urlretrieve(url, filename)
# sample execution (requires torchvision)
from PIL import Image
from torchvision import transforms
input_image = Image.open(filename)
preprocess = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])
input_tensor = preprocess(input_image)
input_batch = input_tensor.unsqueeze(0) # create a mini-batch as expected by the model

# move the input and model to GPU for speed if available
if torch.cuda.is_available():
    input_batch = input_batch.to('cuda')
    model.to('cuda')

with torch.no_grad():
    output = model(input_batch)
# Tensor of shape 1000, with confidence scores over Imagenet's 1000 classes
print(output[0])
# The output has unnormalized scores. To get probabilities, you can run a softmax on it.
probabilities = torch.nn.functional.softmax(output[0], dim=0)
print(probabilities)
# Download ImageNet labels
!wget https://raw.githubusercontent.com/pytorch/hub/master/imagenet_classes.txt

# Read the categories
with open("imagenet_classes.txt", "r") as f:
    categories = [s.strip() for s in f.readlines()]
# Show top categories per image
top5_prob, top5_catid = torch.topk(probabilities, 5)
for i in range(top5_prob.size(0)):
    print(categories[top5_catid[i]], top5_prob[i].item())

Model Description

IBN-Net is a CNN model with domain/appearance invariance. Motivated by style transfer works, IBN-Net carefully unifies instance normalization and batch normalization in a single deep network. It provides a simple way to increase both modeling and generalization capacities without adding model complexity. IBN-Net is especially suitable for cross domain or person/vehicle re-identification tasks.

The corresponding accuracies on ImageNet dataset with pretrained models are listed below.

Model name Top-1 acc Top-5 acc
resnet50_ibn_a 77.46 93.68
resnet101_ibn_a 78.61 94.41
resnext101_ibn_a 79.12 94.58
se_resnet101_ibn_a 78.75 94.49

The rank1/mAP on two Re-ID benchmarks Market1501 and DukeMTMC-reID are listed below (from michuanhaohao/reid-strong-baseline).

Backbone Market1501 DukeMTMC-reID
ResNet50 94.5 (85.9) 86.4 (76.4)
ResNet101 94.5 (87.1) 87.6 (77.6)
SeResNet50 94.4 (86.3) 86.4 (76.5)
SeResNet101 94.6 (87.3) 87.5 (78.0)
SeResNeXt50 94.9 (87.6) 88.0 (78.3)
SeResNeXt101 95.0 (88.0) 88.4 (79.0)
ResNet50-IBN-a 95.0 (88.2) 90.1 (79.1)

References