kerasformers/dino_resnet50

See our collection for all versions of DINO.

Run DINO with Keras 3: JAX, PyTorch, or TensorFlow

kerasformers/dino_resnet50

Paper: Emerging Properties in Self-Supervised Vision Transformers (arXiv:2104.14294) · HF Papers

DINO is self-supervised: a student and teacher match across crops of the same image with no labels. The resulting features are semantic for free. These checkpoints are backbones that return tokens / feature maps.

Pure-Keras 3 port for kerasformers, converted from the official upstream release. One implementation runs unmodified on TensorFlow / Torch / JAX.

This is a self-supervised backbone (DinoResNetModel), not a task head.

✨ Quick start

import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

import numpy as np
from PIL import Image
from kerasformers.models.dino import DinoResNetModel

model = DinoResNetModel.from_weights("kerasformers/dino_resnet50", image_size=448)
image = Image.open("your_image.jpg").convert("RGB")
x = np.asarray(image.resize((448, 448)))[None].astype("float32")
tokens = model(x, training=False)
print(tokens.shape)

Load any DINO variant the same way with from_weights("kerasformers/<variant>"):

Variant	Hub	Backbone
dino_vits16	kerasformers/dino_vits16	ViT-S/16
dino_vits8	kerasformers/dino_vits8	ViT-S/8
dino_vitb16	kerasformers/dino_vitb16	ViT-B/16
dino_vitb8	kerasformers/dino_vitb8	ViT-B/8
dino_resnet50	kerasformers/dino_resnet50	ResNet-50

Tips

• Set KERAS_BACKEND before importing Keras / kerasformers.
• Feed raw [0, 255] pixels; normalization happens inside by default.
• dino_resnet50 was converted from torch.hub facebookresearch/dino.
• See DINO docs and Loading Weights.
• Community / upstream weights: See the KerasFormers docs for upstream conversion notes.

Special Thanks

A huge thank you to the Facebook AI Research DINO authors for creating and releasing these models.

License: Apache 2.0.