Model card for test_nfnet.r160_in1k

A very small test NF-Net image classification model for testing and sanity checks. Trained on ImageNet-1k by Ross Wightman.

Model Details

• Model Type: Image classification / feature backbone
• Model Stats:
  • Params (M): 0.4
  • GMACs: 0.3
  • Activations (M): 1.2
  • Image size: 160 x 160
• Dataset: ImageNet-1k
• Papers:
  • PyTorch Image Models: https://github.com/huggingface/pytorch-image-models
• Original: https://github.com/huggingface/pytorch-image-models

Model Usage

Image Classification

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen(
    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
))

model = timm.create_model('test_nfnet.r160_in1k', pretrained=True)
model = model.eval()

# get model specific transforms (normalization, resize)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0))  # unsqueeze single image into batch of 1

top5_probabilities, top5_class_indices = torch.topk(output.softmax(dim=1) * 100, k=5)

Feature Map Extraction

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen(
    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
))

model = timm.create_model(
    'test_nfnet.r160_in1k',
    pretrained=True,
    features_only=True,
)
model = model.eval()

# get model specific transforms (normalization, resize)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0))  # unsqueeze single image into batch of 1

for o in output:
    # print shape of each feature map in output
    # e.g.:
    #  torch.Size([1, 64, 80, 80])
    #  torch.Size([1, 32, 40, 40])
    #  torch.Size([1, 64, 20, 20])
    #  torch.Size([1, 96, 10, 10])
    #  torch.Size([1, 192, 5, 5])

    print(o.shape)

Image Embeddings

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen(
    'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
))

model = timm.create_model(
    'test_nfnet.r160_in1k',
    pretrained=True,
    num_classes=0,  # remove classifier nn.Linear
)
model = model.eval()

# get model specific transforms (normalization, resize)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0))  # output is (batch_size, num_features) shaped tensor

# or equivalently (without needing to set num_classes=0)

output = model.forward_features(transforms(img).unsqueeze(0))
# output is unpooled, a (1, 192, 5, 5) shaped tensor

output = model.forward_head(output, pre_logits=True)
# output is a (1, num_features) shaped tensor

Model Comparison

By Top-1

model	img_size	top1	top5	param_count
test_convnext3.r160_in1k	192	54.558	79.356	0.47
test_convnext2.r160_in1k	192	53.62	78.636	0.48
test_convnext2.r160_in1k	160	53.51	78.526	0.48
test_convnext3.r160_in1k	160	53.328	78.318	0.47
test_convnext.r160_in1k	192	48.532	74.944	0.27
test_nfnet.r160_in1k	192	48.298	73.446	0.38
test_convnext.r160_in1k	160	47.764	74.152	0.27
test_nfnet.r160_in1k	160	47.616	72.898	0.38
test_efficientnet.r160_in1k	192	47.164	71.706	0.36
test_efficientnet_evos.r160_in1k	192	46.924	71.53	0.36
test_byobnet.r160_in1k	192	46.688	71.668	0.46
test_efficientnet_evos.r160_in1k	160	46.498	71.006	0.36
test_efficientnet.r160_in1k	160	46.454	71.014	0.36
test_byobnet.r160_in1k	160	45.852	70.996	0.46
test_efficientnet_ln.r160_in1k	192	44.538	69.974	0.36
test_efficientnet_gn.r160_in1k	192	44.448	69.75	0.36
test_efficientnet_ln.r160_in1k	160	43.916	69.404	0.36
test_efficientnet_gn.r160_in1k	160	43.88	69.162	0.36
test_vit2.r160_in1k	192	43.454	69.798	0.46
test_resnet.r160_in1k	192	42.376	68.744	0.47
test_vit2.r160_in1k	160	42.232	68.982	0.46
test_vit.r160_in1k	192	41.984	68.64	0.37
test_resnet.r160_in1k	160	41.578	67.956	0.47
test_vit.r160_in1k	160	40.946	67.362	0.37

Citation

@misc{rw2019timm,
  author = {Ross Wightman},
  title = {PyTorch Image Models},
  year = {2019},
  publisher = {GitHub},
  journal = {GitHub repository},
  doi = {10.5281/zenodo.4414861},
  howpublished = {\url{https://github.com/huggingface/pytorch-image-models}}
}