File size: 20,671 Bytes
ad1d2d7
3fad000
ad1d2d7
3fad000
 
ad1d2d7
3fad000
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ad1d2d7
3fad000
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
---
title: DinoV2_Semantic_Segmentation
app_file: app.py
sdk: gradio
sdk_version: 3.42.0
---
# DINOv2: Learning Robust Visual Features without Supervision

**[Meta AI Research, FAIR](https://ai.facebook.com/research/)**

Maxime Oquab,
Timothée Darcet,
Théo Moutakanni,
Huy V. Vo,
Marc Szafraniec,
Vasil Khalidov,
Patrick Labatut,
Armand Joulin,
Piotr Bojanowski

[[`Paper`](https://arxiv.org/abs/2304.07193)] [[`Blog`](https://ai.facebook.com/blog/dino-v2-computer-vision-self-supervised-learning/)] [[`Demo`](https://dinov2.metademolab.com)] [[`BibTeX`](#citing-dinov2)]

PyTorch implementation and pretrained models for DINOv2. For details, see the paper: **[DINOv2: Learning Robust Visual Features without Supervision](https://arxiv.org/abs/2304.07193)**.

DINOv2 models produce high-performance visual features that can be directly employed with classifiers as simple as linear layers on a variety of computer vision tasks; these visual features are robust and perform well across domains without any requirement for fine-tuning. The models were pretrained on a dataset of 142 M images without using any labels or annotations.

https://github.com/facebookresearch/dinov2/assets/60359573/f168823e-7922-415a-b429-578badf5c356

<div align="center">
  Visualization of the three first principal components of the patch features of all frames, mapped to RGB values.
</div>

## Pretrained models

<table style="margin: auto">
  <thead>
    <tr>
      <th>model</th>
      <th># of<br />params</th>
      <th>ImageNet<br />k-NN</th>
      <th>ImageNet<br />linear</th>
      <th>download</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>ViT-S/14 distilled</td>
      <td align="right">21 M</td>
      <td align="right">79.0%</td>
      <td align="right">81.1%</td>
      <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_pretrain.pth">backbone only</a></td>
    </tr>
    <tr>
      <td>ViT-B/14 distilled</td>
      <td align="right">86 M</td>
      <td align="right">82.1%</td>
      <td align="right">84.5%</td>
      <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_pretrain.pth">backbone only</a></td>
    </tr>
    <tr>
      <td>ViT-L/14 distilled</td>
      <td align="right">300 M</td>
      <td align="right">83.5%</td>
      <td align="right">86.3%</td>
      <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_pretrain.pth">backbone only</a></td>
    </tr>
    <tr>
      <td>ViT-g/14</td>
      <td align="right">1,100 M</td>
      <td align="right">83.5%</td>
      <td align="right">86.5%</td>
      <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_pretrain.pth">backbone only</a></td>
    </tr>
  </tbody>
</table>

### Pretrained backbones (via PyTorch Hub)

Please follow the instructions [here](https://pytorch.org/get-started/locally/) to install PyTorch (the only required dependency for loading the model). Installing PyTorch with CUDA support is strongly recommended.

A corresponding [model card](MODEL_CARD.md) is included in the repository.

```python
import torch

dinov2_vits14 = torch.hub.load('facebookresearch/dinov2', 'dinov2_vits14')
dinov2_vitb14 = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitb14')
dinov2_vitl14 = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitl14')
dinov2_vitg14 = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitg14')
```

### Pretrained heads - Image classification

<table style="margin: auto">
  <thead>
    <tr>
      <th rowspan="2">backbone</th>
      <th>download</th>
    </tr>
    <tr>
      <th>ImageNet</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>ViT-S/14 distilled</td>
      <td>
        linear head (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_linear4_head.pth">4 layers</a>)
      </td>
    </tr>
    <tr>
      <td>ViT-B/14 distilled</td>
      <td>
        linear head (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_linear4_head.pth">4 layers</a>)
    </tr>
    <tr>
      <td>ViT-L/14 distilled</td>
      <td>
        linear head (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_linear4_head.pth">4 layers</a>)
    </tr>
    <tr>
      <td>ViT-g/14</td>
      <td>
        linear head (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_linear4_head.pth">4 layers</a>)
    </tr>
  </tbody>
</table>

The (full) classifier models can be loaded via PyTorch Hub:

```python
import torch

dinov2_vits14_lc = torch.hub.load('facebookresearch/dinov2', 'dinov2_vits14_lc')
dinov2_vitb14_lc = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitb14_lc')
dinov2_vitl14_lc = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitl14_lc')
dinov2_vitg14_lc = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitg14_lc')
```

### Pretrained heads - Depth estimation

<table style="margin: auto">
  <thead>
    <tr>
      <th rowspan="2">backbone</th>
      <th colspan="2">download head</th>
    </tr>
    <tr>
      <th>NYUd</th>
      <th>KITTI</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>ViT-S/14 distilled</td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_nyu_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_nyu_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_nyu_dpt_head.pth">DPT</a>
      </td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_kitti_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_kitti_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_kitti_dpt_head.pth">DPT</a>
      </td>
    </tr>
    <tr>
      <td>ViT-B/14 distilled</td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_nyu_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_nyu_dpt_head.pth">DPT</a>
      </td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_kitti_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_kitti_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_kitti_dpt_head.pth">DPT</a>
      </td>
    </tr>
    <tr>
      <td>ViT-L/14 distilled</td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_nyu_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_nyu_dpt_head.pth">DPT</a>
      </td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_kitti_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_kitti_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_kitti_dpt_head.pth">DPT</a>
      </td>
    </tr>
    <tr>
      <td>ViT-g/14</td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_nyu_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_nyu_dpt_head.pth">DPT</a>
      </td>
      <td>
        linear (<a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_kitti_linear_head.pth">1 layer</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_kitti_linear4_head.pth">4 layers</a>),
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_kitti_dpt_head.pth">DPT</a>
      </td>
    </tr>
  </tbody>
</table>

### Pretrained heads - Semantic segmentation

<table style="margin: auto">
  <thead>
    <tr>
      <th rowspan="2">backbone</th>
      <th>download model</th>
      <th colspan="2">download head</th>
    </tr>
    <tr>
      <th>ADE20K</th>
      <th>ADE20K</th>
      <th>VOC2012</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>ViT-S/14 distilled</td>
      <td></td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_ade20k_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_ade20k_ms_head.pth">multi-scale</a>
      </td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_voc2012_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_voc2012_ms_head.pth">multi-scale</a>
      </td>
    </tr>
    <tr>
      <td>ViT-B/14 distilled</td>
      <td></td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_ade20k_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_ade20k_ms_head.pth">multi-scale</a>
      </td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_voc2012_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_voc2012_ms_head.pth">multi-scale</a>
      </td>
    </tr>
    <tr>
      <td>ViT-L/14 distilled</td>
      <td></td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_ade20k_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_ade20k_ms_head.pth">multi-scale</a>
      </td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_voc2012_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_voc2012_ms_head.pth">multi-scale</a>
      </td>
    </tr>
    <tr>
      <td>ViT-g/14</td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_ade20k_m2f.pth">Mask2Former</a>
      </td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_ade20k_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_ade20k_ms_head.pth">multi-scale</a>
      </td>
      <td>
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_voc2012_linear_head.pth">linear</a>,
        <a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_voc2012_ms_head.pth">multi-scale</a>
      </td>
    </tr>
  </tbody>
</table>

## Installation

The training and evaluation code requires PyTorch 2.0 and [xFormers](https://github.com/facebookresearch/xformers) 0.0.18 as well as a number of other 3rd party packages. Note that the code has only been tested with the specified versions and also expects a Linux environment. To setup all the required dependencies for training and evaluation, please follow the instructions below:

*[conda](https://docs.conda.io/projects/conda/en/latest/user-guide/getting-started.html)* **(Recommended)** - Clone the repository and then create and activate a `dinov2` conda environment using the provided environment definition:

```shell
conda env create -f conda.yaml
conda activate dinov2
```

*[pip](https://pip.pypa.io/en/stable/getting-started/)* - Clone the repository and then use the provided `requirements.txt` to install the dependencies:

```shell
pip install -r requirements.txt
```

For dense tasks (depth estimation and semantic segmentation), there are additional dependencies (specific versions of `mmcv` and `mmsegmentation`) which are captured in the `extras` dependency specifications:

*[conda](https://docs.conda.io/projects/conda/en/latest/user-guide/getting-started.html)* **(Recommended)**:

```shell
conda env create -f conda-extras.yaml
conda activate dinov2-extras
```

*[pip](https://pip.pypa.io/en/stable/getting-started/)*:

```shell
pip install -r requirements.txt -r requirements-extras.txt
```

## Data preparation

### ImageNet-1k

The root directory of the dataset should hold the following contents:

- `<ROOT>/test/ILSVRC2012_test_00000001.JPEG`
- `<ROOT>/test/[..]`
- `<ROOT>/test/ILSVRC2012_test_00100000.JPEG`
- `<ROOT>/train/n01440764/n01440764_10026.JPEG`
- `<ROOT>/train/[...]`
- `<ROOT>/train/n15075141/n15075141_9993.JPEG`
- `<ROOT>/val/n01440764/ILSVRC2012_val_00000293.JPEG`
- `<ROOT>/val/[...]`
- `<ROOT>/val/n15075141/ILSVRC2012_val_00049174.JPEG`
- `<ROOT>/labels.txt`

The provided dataset implementation expects a few additional metadata files to be present under the extra directory:

- `<EXTRA>/class-ids-TRAIN.npy`
- `<EXTRA>/class-ids-VAL.npy`
- `<EXTRA>/class-names-TRAIN.npy`
- `<EXTRA>/class-names-VAL.npy`
- `<EXTRA>/entries-TEST.npy`
- `<EXTRA>/entries-TRAIN.npy`
- `<EXTRA>/entries-VAL.npy`

These metadata files can be generated (once) with the following lines of Python code:

```python
from dinov2.data.datasets import ImageNet

for split in ImageNet.Split:
    dataset = ImageNet(split=split, root="<ROOT>", extra="<EXTRA>")
    dataset.dump_extra()
```

Note that the root and extra directories do not have to be distinct directories.

### ImageNet-22k

Please adapt the [dataset class](dinov2/data/datasets/image_net_22k.py) to match your local setup.

<br />

:warning: To execute the commands provided in the next sections for training and evaluation, the `dinov2` package should be included in the Python module search path, i.e. simply prefix the command to run with `PYTHONPATH=.`.

## Training

### Fast setup: training DINOv2 ViT-L/16 on ImageNet-1k

Run DINOv2 training on 4 A100-80GB nodes (32 GPUs) in a SLURM cluster environment with submitit:

```shell
python dinov2/run/train/train.py \
    --nodes 4 \
    --config-file dinov2/configs/train/vitl16_short.yaml \
    --output-dir <PATH/TO/OUTPUT/DIR> \
    train.dataset_path=ImageNet:split=TRAIN:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

Training time is approximately 1 day and the resulting checkpoint should reach 81.6% on k-NN eval and 82.9% on linear eval.

The training code saves the weights of the teacher in the `eval` folder every 12500 iterations for evaluation.

### Long setup: training DINOv2 ViT-L/14 on ImageNet-22k

Run DINOv2 training on 12 A100-80GB nodes (96 GPUs) in a SLURM cluster environment with submitit:

```shell
python dinov2/run/train/train.py \
    --nodes 12 \
    --config-file dinov2/configs/train/vitl14.yaml \
    --output-dir <PATH/TO/OUTPUT/DIR> \
    train.dataset_path=ImageNet22k:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

Training time is approximately 3.3 days and the resulting checkpoint should reach 82.0% on k-NN eval and 84.5% on linear eval.

The training code saves the weights of the teacher in the `eval` folder every 12500 iterations for evaluation.


## Evaluation

The training code regularly saves the teacher weights. In order to evaluate the model, run the following evaluation on a single node:

### k-NN classification on ImageNet-1k

```shell
python dinov2/run/eval/knn.py \
    --config-file <PATH/TO/OUTPUT/DIR>/config.yaml \
    --pretrained-weights <PATH/TO/OUTPUT/DIR>/eval/training_24999/teacher_checkpoint.pth \
    --output-dir <PATH/TO/OUTPUT/DIR>/eval/training_24999/knn \
    --train-dataset ImageNet:split=TRAIN:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET> \
    --val-dataset ImageNet:split=VAL:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

### Logistic regression classification on ImageNet-1k

```shell
python dinov2/run/eval/log_regression.py \
    --config-file <PATH/TO/OUTPUT/DIR>/config.yaml \
    --pretrained-weights <PATH/TO/OUTPUT/DIR>/eval/training_24999/teacher_checkpoint.pth \
    --output-dir <PATH/TO/OUTPUT/DIR>/eval/training_24999/logreg \
    --train-dataset ImageNet:split=TRAIN:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET> \
    --val-dataset ImageNet:split=VAL:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

### Linear classification with data augmentation on ImageNet-1k

```shell
python dinov2/run/eval/linear.py \
    --config-file <PATH/TO/OUTPUT/DIR>/config.yaml \
    --pretrained-weights <PATH/TO/OUTPUT/DIR>/eval/training_24999/teacher_checkpoint.pth \
    --output-dir <PATH/TO/OUTPUT/DIR>/eval/training_24999/linear \
    --train-dataset ImageNet:split=TRAIN:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET> \
    --val-dataset ImageNet:split=VAL:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

We release the weights from evaluating the different models:

<table style="margin: auto">
  <tr>
    <th>model</th>
    <th>ImageNet<br />top-1</th>
    <th>linear evaluation</th>
  </tr>
  <tr>
    <td>ViT-S/14 distilled</td>
    <td align="right">81.1%</td>
    <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vits14/dinov2_vits14_linear_head.pth">linear head weights</a></td>
  </tr>
  <tr>
    <td>ViT-B/14 distilled</td>
    <td align="right">84.5%</td>
    <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitb14/dinov2_vitb14_linear_head.pth">linear head weights</a></td>
  </tr>
  <tr>
    <td>ViT-L/14 distilled</td>
    <td align="right">86.3%</td>
    <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitl14/dinov2_vitl14_linear_head.pth">linear head weights</a></td>
  </tr>
  <tr>
    <td>ViT-g/14</td>
    <td align="right">86.5%</td>
    <td><a href="https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_linear_head.pth">linear head weights</a></td>
  </tr>
</table>

The performance of the provided pretrained model weights can be evaluated as follows on ImageNet-1k:

```shell
python dinov2/run/eval/linear.py \
    --config-file dinov2/configs/eval/vitg14_pretrain.yaml \
    --pretrained-weights https://dl.fbaipublicfiles.com/dinov2/dinov2_vitg14/dinov2_vitg14_pretrain.pth \
    --train-dataset ImageNet:split=TRAIN:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET> \
    --val-dataset ImageNet:split=VAL:root=<PATH/TO/DATASET>:extra=<PATH/TO/DATASET>
```

## Notebooks

A few notebooks are provided to help the community leverage the models and code:

<ul>
  <li><a href="https://github.com/facebookresearch/dinov2/blob/main/notebooks/depth_estimation.ipynb">Depth estimation</a> - How to load and use the depth heads in combination with a matching backbone via mmcv</li>
  <li><a href="https://github.com/facebookresearch/dinov2/blob/main/notebooks/semantic_segmentation.ipynb">Semantic segmentation</a> - How to load and use the segmentation heads in combination with a matching backbone via mmcv, and also how to load and use the Mask2Former-based segmentation model trained on ADE20K</li>
</ul>

## License

DINOv2 code and model weights are released under the Apache License 2.0. See [LICENSE](LICENSE) for additional details.

## Contributing

See [contributing](CONTRIBUTING.md) and the [code of conduct](CODE_OF_CONDUCT.md).

## Citing DINOv2

If you find this repository useful, please consider giving a star :star: and citation :t-rex::

```
@misc{oquab2023dinov2,
  title={DINOv2: Learning Robust Visual Features without Supervision},
  author={Oquab, Maxime and Darcet, Timothée and Moutakanni, Theo and Vo, Huy V. and Szafraniec, Marc and Khalidov, Vasil and Fernandez, Pierre and Haziza, Daniel and Massa, Francisco and El-Nouby, Alaaeldin and Howes, Russell and Huang, Po-Yao and Xu, Hu and Sharma, Vasu and Li, Shang-Wen and Galuba, Wojciech and Rabbat, Mike and Assran, Mido and Ballas, Nicolas and Synnaeve, Gabriel and Misra, Ishan and Jegou, Herve and Mairal, Julien and Labatut, Patrick and Joulin, Armand and Bojanowski, Piotr},
  journal={arXiv:2304.07193},
  year={2023}
}
```