Skip to content

Commit

Permalink
log
Browse files Browse the repository at this point in the history
  • Loading branch information
Wangjii committed May 1, 2023
1 parent 173e565 commit 0f0fec5
Show file tree
Hide file tree
Showing 9 changed files with 1,181 additions and 18 deletions.
2 changes: 0 additions & 2 deletions .gitignore
Original file line number Diff line number Diff line change
@@ -1,4 +1,2 @@
*.pyc
runs/
logs/
.vscode/
61 changes: 45 additions & 16 deletions README.md
Original file line number Diff line number Diff line change
@@ -1,21 +1,50 @@

## 运行
# StyleEDL: Style-Guided High-order Attention Network for Image
## Abstract
the tendency to express emotions through images. As for emotion
ambiguity arising from humans’ subjectivity, substantial previous
methods generally focused on learning appropriate representations
from the holistic or significant part of images. However, they rarely
consider establishing connections with the stylistic information although it can lead to a better understanding of images. In this paper,
we propose a style-guided high-order attention network for image
emotion distribution learning termed StyleEDL, which interactively
learns stylistic-aware representations of images by exploring the
hierarchical stylistic information of visual contents. Specifically, we
consider exploring the intra- and inter-layer correlations among
GRAM-based stylistic representations, and meanwhile exploit an
adversary-constrained high-order attention mechanism to capture
potential interactions between subtle visual parts. In addition, we
introduce a stylistic graph convolutional network to dynamically
generate the content-dependent emotion representations to benefit the final emotion distribution learning. Extensive experiments
conducted on several benchmark datasets demonstrate the effectiveness of our proposed StyleEDL compared to state-of-the-art
methods.

## Method
![alt](./docs/framwork.png)

## Results
### Twitter-LDL ([log](logs/Twitter_LDL/train.log))
![twitter](docs/twitter.png)

### Emotion6 ([log](logs/Emotion6/train.log))
![emotion](docs/emotion.png)

### Flickr-LDL ([log](logs/Flickr_LDL/train.log))
![flickr](docs/flickr.png)

## Installation
```bash
python main.py --tag
conda env create -f env.yaml
```
+ `--tag` 指定保存目录的前缀,默认为cache

## train
```bash
python main.py --tag xxx
```
+ `--tag` the log will saved at `logs/$datetime.time$xxx`


## 添加新的模型

1.`network`文件夹下新建新的文件
1.`network``__init__`中的`models`加入新的网络模型
1.`configs`文件夹中新建对应的配置文件


## 添加新的trainer

1.`trainer`文件夹中新建文件,继承`basetrainer`
1.`trainer``__init__`中的`trainers`加入新的trainer
1. 运行时,由配置文件中的`trainer`指定
## Citation
If you find this repository helpful, please consider citing:
```
add latter.
```
Binary file added docs/emotion.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/flickr.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/framwork.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/twitter.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
360 changes: 360 additions & 0 deletions logs/Emotion6/train.log

Large diffs are not rendered by default.

360 changes: 360 additions & 0 deletions logs/Flickr_LDL/train.log

Large diffs are not rendered by default.

416 changes: 416 additions & 0 deletions logs/Twitter_LDL/train.log

Large diffs are not rendered by default.

0 comments on commit 0f0fec5

Please sign in to comment.