Downloads · 30 days
15
4% of all-time downloads
birder-project/vit_reg4_m16_rms_avg_i-jepa-imagenet21k
vit_reg4_m16_rms_avg_i-jepa-imagenet21k is a image feature extraction model from birder-project. Use it for the image feature extraction task on the model card, and read the license before you ship it in a product. It is set up for birder. The card lists the license as apache-2.0.
A ViT image classification model. The model follows a two-stage training process: first, I-JEPA pretraining, then fine-tuned on the ImageNet-21K dataset.
Downloads · 30 days
15
4% of all-time downloads
All-time downloads
379
Public
Repo size
579 MB
Likes
0
Public
Click a slice to open those files.
.pt193 MB · 100%
From the Hugging Face model README
A ViT image classification model. The model follows a two-stage training process: first, I-JEPA pretraining, then fine-tuned on the ImageNet-21K dataset.
Model Type: Image classification and detection backbone
Model Stats:
Dataset: ImageNet-21K (19167 classes)
Papers:
import birder
from birder.inference.classification import infer_image
(net, model_info) = birder.load_pretrained_model("vit_reg4_m16_rms_avg_i-jepa-imagenet21k", inference=True)
# Get the image size the model was trained on
size = birder.get_size_from_signature(model_info.signature)
# Create an inference transform
transform = birder.classification_transform(size, model_info.rgb_stats)
image = "path/to/image.jpeg" # or a PIL image, must be loaded in RGB format
(out, _) = infer_image(net, image, transform)
# out is a NumPy array with shape of (1, 19167), representing class probabilities.
import birder
from birder.inference.classification import infer_image
(net, model_info) = birder.load_pretrained_model("vit_reg4_m16_rms_avg_i-jepa-imagenet21k", inference=True)
# Get the image size the model was trained on
size = birder.get_size_from_signature(model_info.signature)
# Create an inference transform
transform = birder.classification_transform(size, model_info.rgb_stats)
image = "path/to/image.jpeg" # or a PIL image
(out, embedding) = infer_image(net, image, transform, return_embedding=True)
# embedding is a NumPy array with shape of (1, 512)
from PIL import Image
import birder
(net, model_info) = birder.load_pretrained_model("vit_reg4_m16_rms_avg_i-jepa-imagenet21k", inference=True)
# Get the image size the model was trained on
size = birder.get_size_from_signature(model_info.signature)
# Create an inference transform
transform = birder.classification_transform(size, model_info.rgb_stats)
image = Image.open("path/to/image.jpeg")
features = net.detection_features(transform(image).unsqueeze(0))
# features is a dict (stage name -> torch.Tensor)
print([(k, v.size()) for k, v in features.items()])
# Output example:
# [('neck', torch.Size([1, 512, 16, 16]))]
@misc{dosovitskiy2021imageworth16x16words,
title={An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale},
author={Alexey Dosovitskiy and Lucas Beyer and Alexander Kolesnikov and Dirk Weissenborn and Xiaohua Zhai and Thomas Unterthiner and Mostafa Dehghani and Matthias Minderer and Georg Heigold and Sylvain Gelly and Jakob Uszkoreit and Neil Houlsby},
year={2021},
eprint={2010.11929},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2010.11929},
}
@misc{darcet2024visiontransformersneedregisters,
title={Vision Transformers Need Registers},
author={Timothée Darcet and Maxime Oquab and Julien Mairal and Piotr Bojanowski},
year={2024},
eprint={2309.16588},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2309.16588},
}
@misc{assran2023selfsupervisedlearningimagesjointembedding,
title={Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture},
author={Mahmoud Assran and Quentin Duval and Ishan Misra and Piotr Bojanowski and Pascal Vincent and Michael Rabbat and Yann LeCun and Nicolas Ballas},
year={2023},
eprint={2301.08243},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2301.08243},
}