Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix typo #122

Open
wants to merge 10 commits into
base: master
Choose a base branch
from
7 changes: 4 additions & 3 deletions facebookresearch_pytorchvideo_resnet.md
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ demo-model-link: https://huggingface.co/spaces/pytorch/3D_ResNet

### 사용 예시

#### Imports
#### 불러오기

모델 불러오기:

Expand Down Expand Up @@ -47,7 +47,7 @@ from pytorchvideo.transforms import (
)
```

#### 환경설정
#### 셋업

모델을 평가 모드로 설정하고 원하는 디바이스 방식을 선택합니다.

Expand All @@ -58,7 +58,7 @@ model = model.eval()
model = model.to(device)
```

토치 허브 모델이 훈련된 Kinetics 400 데이터셋에 대해 ID에서의 레이블과 맞는 정보를 다운로드합니다. 이는 예측된 클래스 ID에서 카테고리 레이블 이름을 가져오는데 사용됩니다.
토치 허브 모델이 훈련된 Kinetics 400 데이터셋에 대해 ID에서의 레이블 매핑 정보를 다운로드합니다. 이는 예측된 클래스 ID에서 카테고리 레이블 이름을 가져오는데 사용됩니다.

```python
json_url = "https://dl.fbaipublicfiles.com/pyslowfast/dataset/class_names/kinetics_classnames.json"
Expand Down Expand Up @@ -159,6 +159,7 @@ print("Top 5 predicted labels: %s" % ", ".join(pred_class_names))

### 모델 설명
모델 아키텍처는 Kinetics 데이터셋의 8x8 설정을 사용하여 사전 훈련된 가중치가 있는 참고문헌 [1]을 기반으로 합니다.

| arch | depth | frame length x sample rate | top 1 | top 5 | Flops (G) | Params (M) |
| --------------- | ----------- | ----------- | ----------- | ----------- | ----------- | ----------- |
| Slow | R50 | 8x8 | 74.58 | 91.63 | 54.52 | 32.45 |
Expand Down
2 changes: 1 addition & 1 deletion facebookresearch_pytorchvideo_slowfast.md
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,7 @@ for k, v in kinetics_classnames.items():
kinetics_id_to_classname[v] = str(k).replace('"', "")
```

#### 입력 변환에 대한 정의
#### 입력 형태에 대한 정의

```python
side_size = 256
Expand Down
2 changes: 1 addition & 1 deletion facebookresearch_pytorchvideo_x3d.md
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ demo-model-link: https://huggingface.co/spaces/pytorch/X3D

### 사용 예시

#### Imports
#### 불러오기

모델 불러오기:

Expand Down
16 changes: 12 additions & 4 deletions hustvl_yolop.md
Original file line number Diff line number Diff line change
Expand Up @@ -125,7 +125,15 @@ det_out, da_seg_out,ll_seg_out = model(img)

### 인용(Citation)

See for more detail in [github code](https://github.com/hustvl/YOLOP) and [arxiv paper](https://arxiv.org/abs/2108.11250).

본 논문과 코드가 여러분의 연구에 유용하다고 판단되면, GitHub star를 주는 것과 본 논문을 인용하는 것을 고려해 주세요:

본 [논문](https://arxiv.org/abs/2108.11250) 과 [코드](https://github.com/hustvl/YOLOP) 가 여러분의 연구에 유용하다고 판단되면, GitHub star를 주는 것과 본 논문을 인용하는 것을 고려해 주세요:

```BibTeX
@article{wu2022yolop,
title={Yolop: You only look once for panoptic driving perception},
author={Wu, Dong and Liao, Man-Wen and Zhang, Wei-Tian and Wang, Xing-Gang and Bai, Xiang and Cheng, Wen-Qing and Liu, Wen-Yu},
journal={Machine Intelligence Research},
pages={1--13},
year={2022},
publisher={Springer}
}
```
2 changes: 1 addition & 1 deletion pytorch_vision_resnext.md
Original file line number Diff line number Diff line change
Expand Up @@ -85,7 +85,7 @@ for i in range(top5_prob.size(0)):

### 모델 설명

Resnext 모델은 논문 [Aggregated Residual Transformations for Deep Neural Networks]에서 제안되었습니다. (https://arxiv.org/abs/1611.05431).
Resnext 모델은 논문 ["Aggregated Residual Transformations for Deep Neural Networks"](https://arxiv.org/abs/1611.05431) 에서 제안되었습니다.
여기서는 50개의 계층과 101개의 계층을 가지는 2개의 resnet 모델을 제공하고 있습니다.
resnet50과 resnext50의 아키텍처 차이는 논문의 Table 1을 참고하십시오.
ImageNet 데이터셋에 대한 사전훈련된 모델의 에러(성능)은 아래 표와 같습니다.
Expand Down
4 changes: 2 additions & 2 deletions pytorch_vision_vgg.md
Original file line number Diff line number Diff line change
Expand Up @@ -90,12 +90,12 @@ for i in range(top5_prob.size(0)):

### 모델 설명

각 구성 및 bachnorm 버전에 대해서 [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://arxiv.org/abs/1409.1556)에서 제안한 모델에 대한 구현이 있습니다.
각 구성 및 batchnorm 버전에 대해서 [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://arxiv.org/abs/1409.1556)에서 제안한 모델에 대한 구현이 있습니다.

예를 들어, 논문에 제시된 구성 `A`는 `vgg11`, `B`는 `vgg13`, `D`는 `vgg16`, `E`는 `vgg19`입니다.
batchnorm 버전은 `_bn`이 접미사로 붙어있습니다.

사전 훈련된 모델이 있는 imagenet 데이터 세트의 1-crop 오류율은 아래에 나열되어 있습니다.
사전 훈련된 모델이 있는 ImageNet 데이터 세트의 Top-1 오류율은 아래에 나열되어 있습니다.

| Model structure | Top-1 error | Top-5 error |
| --------------- | ----------- | ----------- |
Expand Down
1 change: 1 addition & 0 deletions pytorch_vision_wide_resnet.md
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,7 @@ Wide Residual 네트워크는 ResNet에 비해 단순히 채널 수가 증가했

`wide_resnet50_2` 및 `wide_resnet101_2` 모델은 [Warm Restarts가 있는 SGD(SGDR)](https://arxiv.org/abs/1608.03983)를 사용하여 혼합 정밀도(Mixed Precision) 방식으로 학습되었습니다.
체크 포인트는 크기가 작은 경우 절반 정밀도(batch norm 제외)의 가중치를 가지며 FP32 모델에서도 사용할 수 있습니다.

| Model structure | Top-1 error | Top-5 error | # parameters |
| ----------------- | :---------: | :---------: | :----------: |
| wide_resnet50_2 | 21.49 | 5.91 | 68.9M |
Expand Down