admin
commited on
Commit
·
3482f7b
1
Parent(s):
f952648
Update README.md
Browse files
README.md
CHANGED
@@ -16,8 +16,6 @@ size_categories:
|
|
16 |
This repository consolidates the collection of backbone networks for pre-trained computer vision models available on the PyTorch official website. It mainly includes various Convolutional Neural Networks (CNNs) and Vision Transformer models pre-trained on the ImageNet1K dataset. The entire collection is divided into two subsets, V1 and V2, encompassing multiple classic and advanced versions of visual models. These pre-trained backbone networks provide users with a robust foundation for transfer learning in tasks such as image recognition, object detection, and image segmentation. Simultaneously, it offers a convenient choice for researchers and practitioners to flexibly apply these pre-trained models in different scenarios.
|
17 |
|
18 |
## Viewer
|
19 |
-
<https://huggingface.co/spaces/monetjoe/cv-backbones>
|
20 |
-
### Data Fields
|
21 |
| ver | type | input_size | url |
|
22 |
| :-----------: | :-----------: | :--------------: | :-------------------------------: |
|
23 |
| backbone name | backbone type | input image size | url of pretrained model .pth file |
|
@@ -25,9 +23,11 @@ This repository consolidates the collection of backbone networks for pre-trained
|
|
25 |
## Maintenance
|
26 |
```bash
|
27 |
git clone git@hf.co:datasets/monetjoe/cv_backbones
|
|
|
28 |
```
|
29 |
|
30 |
## Usage
|
|
|
31 |
```python
|
32 |
from datasets import load_dataset
|
33 |
|
@@ -36,6 +36,15 @@ for weights in backbones:
|
|
36 |
print(weights)
|
37 |
```
|
38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
## Param counts of different backbones
|
40 |
### IMAGENET1K_V1
|
41 |
| Backbone | Params(M) |
|
|
|
16 |
This repository consolidates the collection of backbone networks for pre-trained computer vision models available on the PyTorch official website. It mainly includes various Convolutional Neural Networks (CNNs) and Vision Transformer models pre-trained on the ImageNet1K dataset. The entire collection is divided into two subsets, V1 and V2, encompassing multiple classic and advanced versions of visual models. These pre-trained backbone networks provide users with a robust foundation for transfer learning in tasks such as image recognition, object detection, and image segmentation. Simultaneously, it offers a convenient choice for researchers and practitioners to flexibly apply these pre-trained models in different scenarios.
|
17 |
|
18 |
## Viewer
|
|
|
|
|
19 |
| ver | type | input_size | url |
|
20 |
| :-----------: | :-----------: | :--------------: | :-------------------------------: |
|
21 |
| backbone name | backbone type | input image size | url of pretrained model .pth file |
|
|
|
23 |
## Maintenance
|
24 |
```bash
|
25 |
git clone git@hf.co:datasets/monetjoe/cv_backbones
|
26 |
+
cd cv_backbones
|
27 |
```
|
28 |
|
29 |
## Usage
|
30 |
+
### ImageNet V1
|
31 |
```python
|
32 |
from datasets import load_dataset
|
33 |
|
|
|
36 |
print(weights)
|
37 |
```
|
38 |
|
39 |
+
### ImageNet V2
|
40 |
+
```python
|
41 |
+
from datasets import load_dataset
|
42 |
+
|
43 |
+
backbones = load_dataset("monetjoe/cv_backbones", name="default", split="test")
|
44 |
+
for weights in backbones:
|
45 |
+
print(weights)
|
46 |
+
```
|
47 |
+
|
48 |
## Param counts of different backbones
|
49 |
### IMAGENET1K_V1
|
50 |
| Backbone | Params(M) |
|