Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -28,14 +28,19 @@ We provide masks for the following datasets: [ImageNet](https://image-net.org/ch
|
|
| 28 |
|
| 29 |
If you only want to download the mask data (do not contain raw images), please download from [[_OPTIONAL_FOREGROUND_MASK_ONLY_DATA folder](https://huggingface.co/datasets/JREion/Prompt_Tuning_Datasets_with_Foreground/tree/main/_OPTIONAL_FOREGROUND_MASK_ONLY_DATA)].
|
| 30 |
|
|
|
|
|
|
|
| 31 |
# 🏷️ Scope of Application
|
| 32 |
|
| 33 |
Datasets are suitable for training and improving **foreground-supervised prompt tuning** methods. For example:
|
| 34 |
|
| 35 |
-
- _FVG-PT: Adaptive Foreground View-Guided Prompt Tuning for Vision-Language Models_
|
|
|
|
| 36 |
|
| 37 |
Also, they are **fully compatible** with other original prompt tuning approaches.
|
| 38 |
|
|
|
|
|
|
|
| 39 |
# ⚙ Data Preparation
|
| 40 |
|
| 41 |
The datasets include the original images, the `split_zhou_xxx.json` annotations, and **foreground masks**.
|
|
@@ -49,6 +54,8 @@ Additionally, you can prepare ImageNet dataset from: [[Raw Images](https://www.k
|
|
| 49 |
|
| 50 |
**_NOTE: You can build the file tree by referring to the [[FVG-PT repository](https://github.com/JREion/FVG-PT/blob/main/docs/DATASETS.md)]._**
|
| 51 |
|
|
|
|
|
|
|
| 52 |
# 🖥︎ When You Write Your Own Code ...
|
| 53 |
|
| 54 |
1. You can refer to `./Dassl.pytorch` in the [FVG-PT repository](https://github.com/JREion/FVG-PT/tree/main) to build a DataLoader that can pass masks.
|
|
@@ -70,7 +77,7 @@ Additionally, you can prepare ImageNet dataset from: [[Raw Images](https://www.k
|
|
| 70 |
```
|
| 71 |
|
| 72 |
|
| 73 |
-
|
| 74 |
|
| 75 |
# Acknowledgements
|
| 76 |
|
|
|
|
| 28 |
|
| 29 |
If you only want to download the mask data (do not contain raw images), please download from [[_OPTIONAL_FOREGROUND_MASK_ONLY_DATA folder](https://huggingface.co/datasets/JREion/Prompt_Tuning_Datasets_with_Foreground/tree/main/_OPTIONAL_FOREGROUND_MASK_ONLY_DATA)].
|
| 30 |
|
| 31 |
+
<br>
|
| 32 |
+
|
| 33 |
# 🏷️ Scope of Application
|
| 34 |
|
| 35 |
Datasets are suitable for training and improving **foreground-supervised prompt tuning** methods. For example:
|
| 36 |
|
| 37 |
+
- _FVG-PT: Adaptive Foreground View-Guided Prompt Tuning for Vision-Language Models_   [[GitHub](https://github.com/JREion/FVG-PT)]
|
| 38 |
+
|
| 39 |
|
| 40 |
Also, they are **fully compatible** with other original prompt tuning approaches.
|
| 41 |
|
| 42 |
+
<br>
|
| 43 |
+
|
| 44 |
# ⚙ Data Preparation
|
| 45 |
|
| 46 |
The datasets include the original images, the `split_zhou_xxx.json` annotations, and **foreground masks**.
|
|
|
|
| 54 |
|
| 55 |
**_NOTE: You can build the file tree by referring to the [[FVG-PT repository](https://github.com/JREion/FVG-PT/blob/main/docs/DATASETS.md)]._**
|
| 56 |
|
| 57 |
+
<br>
|
| 58 |
+
|
| 59 |
# 🖥︎ When You Write Your Own Code ...
|
| 60 |
|
| 61 |
1. You can refer to `./Dassl.pytorch` in the [FVG-PT repository](https://github.com/JREion/FVG-PT/tree/main) to build a DataLoader that can pass masks.
|
|
|
|
| 77 |
```
|
| 78 |
|
| 79 |
|
| 80 |
+
<br>
|
| 81 |
|
| 82 |
# Acknowledgements
|
| 83 |
|