README.md

# MaskedDenoising
## 论文
[Images Speak in Images: A Generalist Painter for In-Context Visual Learning](https://arxiv.org/abs/2212.02499)

## 模型结构



<div align=center>
    <img src="./doc/method.jpg"/>
</div>

## 算法原理



<div align=center>
    <img src="./doc/progress.png"/>
</div>

## 环境配置

### Docker（方法一）

-v 路径、docker_name和imageID根据实际情况修改

```image.sourcefind.cn:5000/dcu/admin/base/pytorch:1.13.1-centos7.6-dtk-23.04.1-py38-latest
docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:1.13.1-centos7.6-dtk-23.04.1-py38-latest
docker run -it -v /path/your_code_data/:/path/ your_code_data/ --shm-size=32G --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video --name docker_name imageID bash

cd /your_code_path/maskeddenoising_pytorch
pip install --upgrade setuptools wheel
pip install -r requirement.txt
```

### Dockerfile（方法二）

-v 路径、docker_name和imageID根据实际情况修改

```
cd ./docker
cp ../requirement.txt requirement.txt
docker build --no-cache -t maskeddenoising:latest .
docker run -it -v /path/your_code_data/:/path/your_code_data/ --shm-size=32G --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video --name docker_name imageID bash

cd /your_code_path/maskeddenoising_pytorch
pip install --upgrade setuptools wheel
pip install -r requirement.txt
```

### Anaconda（方法三）

1、关于本项目DCU显卡所需的特殊深度学习库可从光合开发者社区下载安装： https://developer.hpccube.com/tool/

```
DTK软件栈：dtk23.04.1
python：python3.8
torch：1.13.1
torchvision：0.14.1
```

Tips：以上dtk软件栈、python、torch等DCU相关工具版本需要严格一一对应

2、其他非特殊库直接按照requirement.txt安装

```bash
pip install --upgrade setuptools wheel
pip install -r requirement.txt
```

## 数据集

### 数据集所需环境配置
#### ADE20K Semantic Segmentation

```bash
git clone https://github.com/facebookresearch/detectron2
python -m pip install -e detectron2
```

#### COCO Panoptic Segmentation

```bash
pip install openmim #（0.3.9）
mim install mmcv-full # 注意版本是不是1.7.1
pip install mmdet==2.26.0 # 对应 mmcv-1.7.1
pip install yapf==0.40.1
```

#### COCO Pose Estimation

pip install mmcv==1.3.9
pip install mmpose==0.29.0

或者也可以直接采用源码安装mmpose
```bash
# choose commit id `8c58a18b`
git clone https://github.com/open-mmlab/mmpose.git
cd mmpose
pip install -r requirements.txt
pip install -v -e .
```

### 数据集下载
项目数据集需求较多, 可以使用提供的[a toy training dataset](https://huggingface.co/BAAI/Painter/blob/main/toy_datasets.tar)数据集来验证功能, 数据集由每个类别中各10个类别组成. 将数据集放置于 `$Painter_ROOT/toy_datasets` 路径下, 并设置`$Painter_ROOT/train_painter_vit_large.sh` 中 `DATA_PATH=toy_datasets`.

完整所需数据集如下所示:

#### NYU Depth V2

首先, 下载数据集[here](https://drive.google.com/file/d/1AysroWpfISmm-yRFGBgFTrLy6FjQwvwP/view?usp=sharing). 确保将下载的数据集存放到 `$Painter_ROOT/datasets/nyu_depth_v2/sync.zip`

接下来准备NYU_Depth_V2测试集[NYU Depth V2 test](https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html).

```bash
# 下载原始 NYU Depth V2 split file
wget -P datasets/nyu_depth_v2/ http://horatio.cs.nyu.edu/mit/silberman/nyu_depth_v2/nyu_depth_v2_labeled.mat
# 将 mat 数据转换成 image files
python data/depth/extract_official_train_test_set_from_mat.py datasets/nyu_depth_v2/nyu_depth_v2_labeled.mat data/depth/splits.mat datasets/nyu_depth_v2/official_splits/
```

最后, 准备训练和验证所需json数据, 生成的json数据将会默认保存到 `$Painter_ROOT/datasets/nyu_depth_v2/` 路径下.

```bash
python data/depth/gen_json_nyuv2_depth.py --split sync
python data/depth/gen_json_nyuv2_depth.py --split test
```

#### ADE20k Semantic Segmentation

首先, 下载数据集 [official website](https://groups.csail.mit.edu/vision/datasets/ADE20K/), 将下载的数据集存放到 `$Painter_ROOT/datasets/`.

接下来, 解压 zip 文件并重命名为`ade20k`. 完成后的 ade20k 文件结构如下所示:
```bash
ade20k/
    images/
    annotations/
```

第二, 执行下面的命令准备训练和验证所需的 annotations, 生成的 annotations 将会默认保存到 `$Painter_ROOT/datasets/ade20k/annotations_with_color/` 路径下.
```bash
python data/ade20k/gen_color_ade20k_sem.py --split training
python data/ade20k/gen_color_ade20k_sem.py --split validation
```

第三, 准备训练和验证所需json文件, 生成的json数据将会默认保存到 `$Painter_ROOT/datasets/ade20k/` 路径下.
```bash
python data/ade20k/gen_json_ade20k_sem.py --split training
python data/ade20k/gen_json_ade20k_sem.py --split validation
```

最后, 为了确认能通过 detectron2 进行验证, 创建 `$Painter_ROOT/datasets/ade20k` to `$Painter_ROOT/datasets/ADEChallengeData2016` 的软连接, 然后执行下面的操作:
```bash
# 创建软连接
# ln -s $Painter_ROOT/datasets/ade20k datasets/ADEChallengeData2016
# 执行
python data/prepare_ade20k_sem_seg.py
```

#### COCO Panoptic Segmentation
下载 COCO2017 数据 和 the corresponding panoptic segmentation annotation. 完成后的 COCO 文件结构如下所示:
```
coco/
    train2017/
    val2017/
    annotations/
        instances_train2017.json
        instances_val2017.json
        panoptic_train2017.json
        panoptic_val2017.json
        panoptic_train2017/
        panoptic_val2017/
```

1. 准备 COCO Semantic Segmentation
准备训练所需的annotations, 生成的annotations默认保存到 `$Painter_ROOT/datasets/coco/pano_sem_seg/` 路径下.
```bash
python data/coco_semseg/gen_color_coco_panoptic_segm.py --split train2017
python data/coco_semseg/gen_color_coco_panoptic_segm.py --split val2017
```

准备训练和验证所需的json数据, 生成的json数据默认保存到 `$Painter_ROOT/datasets/coco/pano_sem_seg/` 路径下.
```bash
python data/coco_semseg/gen_json_coco_panoptic_segm.py --split train2017
python data/coco_semseg/gen_json_coco_panoptic_segm.py --split val2017
```

2. 准备 COCO Class-Agnostic Instance Segmentation

第一步, 通过下面的命令对数据进行预处理, 生成的 painted ground truth 将会默认保存到 `$Painter_ROOT/datasets/coco/pano_ca_inst` 路径下.

```bash
cd $Painter_ROOT/data/mmdet_custom

# 为实例分割生成使用通用数据增强的训练数据, 注意我们通过在configs/coco_panoptic_ca_inst_gen_augg.py中交替生成30个副本train_aug{idx}
./tools/dist_train.sh configs/coco_panoptic_ca_inst_gen_aug.py 1
# 仅使用水平翻转增强生成训练数据
./tools/dist_train.sh configs/coco_panoptic_ca_inst_gen_orgflip.py 1
# 生成无数据增强的训练数据
./tools/dist_train.sh configs/coco_panoptic_ca_inst_gen_org.py 1
# 生成验证数据(无数据增强)
./tools/dist_test.sh configs/coco_panoptic_ca_inst_gen_org.py none 1 --eval segm
```

然后, 准备训练和验证所需json文件. 生成的json文件将会默认保存到 `$Painter_ROOT/datasets/coco/pano_ca_inst` 路径下.

```bash
cd $Painter_ROOT
python data/mmdet_custom/gen_json_coco_panoptic_inst.py --split train
python data/mmdet_custom/gen_json_coco_panoptic_inst.py --split val
```

最后, 为了确保使用detectron2进行验证, 创建`$Painter_ROOT/datasets/coco/annotations/panoptic_val2017` to `$Painter_ROOT/datasets/coco/panoptic_val2017` 的软连接并运行:
```bash
# 创建软连接
# ln -s $Painter_ROOT/datasets/coco/annotations/panoptic_val2017 datasets/coco/panoptic_val2017
# 执行
python data/prepare_coco_semantic_annos_from_panoptic_annos.py
```

#### COCO Human Pose Estimation

首先, 下载COCO val2017的行人检测结果 [google drive](https://drive.google.com/drive/folders/1fRUDNUDxe9fjqcRZ2bnF_TKMlO0nB_dk), 将下载的数据放入 `$Painter_ROOT/datasets/coco_pose/` 路径下

然后, 通过下面的命令对数据进行预处理, 得到的 painted ground truth 默认保存到 `$Painter_ROOT/datasets/coco_pose/` 路径下.

```bash
cd $Painter_ROOT/data/mmpose_custom

# 生成用于姿态估计的通用数据增强的训练数据, 本项目生成20个副本用于训练, 需要对coco_256x192_gendata.py中52行的aug_idx参数进行对应数量修改,当前默认为0
./tools/dist_train.sh configs/coco_256x192_gendata.py 1
# 生成训练期间验证的数据
./tools/dist_test.sh configs/coco_256x192_gendata.py none 1
# 生成用于测试的数据(使用离线盒子)
./tools/dist_test.sh configs/coco_256x192_gendata_test.py none 1
# 生成用于测试的数据(使用离线盒子+翻转)
./tools/dist_test.sh configs/coco_256x192_gendata_testflip.py none 1
```

接着, 准备训练和验证所需json文件. 生成的json文件将会默认保存到 `datasets/pano_ca_inst/` 路径下.
```bash
cd $Painter_ROOT
python data/mmpose_custom/gen_json_coco_pose.py --split train
python data/mmpose_custom/gen_json_coco_pose.py --split val
```

#### Low-level Vision Tasks

##### Deraining
参考[MPRNet](https://github.com/swz30/MPRNet) 进行deraining的数据准备.

跟随[MPRNet](https://github.com/swz30/MPRNet/blob/main/Deraining/Datasets/README.md)的指令步骤下载数据集, 将下载的数据集保存到 `$Painter_ROOT/datasets/derain/`. 完成后的 Derain 文件结构如下所示:
```
derain/
    train/
        input/
        target/
    test/
        Rain100H/
        Rain100L/
        Test100/
        Test1200/
        Test2800/
```

接着, 通过下面的命令, 准备训练和验证所需json文件. 生成的json文件将保存到 `datasets/derain/` 路径下.
```bash
python data/derain/gen_json_rain.py --split train
python data/derain/gen_json_rain.py --split val
```

### Denoising
参考[Uformer](https://github.com/ZhendongWang6/Uformer)准备SIDD denoising数据集.

针对训练用的SIDD数据集, 可从[official url](https://www.eecs.yorku.ca/~kamel/sidd/dataset.php)中下载SIDD-Medium dataset数据;

针对验证用的SIDD数据集. 可以从[here](https://mailustceducn-my.sharepoint.com/:f:/g/personal/zhendongwang_mail_ustc_edu_cn/Ev832uKaw2JJhwROKqiXGfMBttyFko_zrDVzfSbFFDoi4Q?e=S3p5hQ)下载.

接下来, 使用以下命令生成用于训练的图像补丁:
```bash
python data/sidd/generate_patches_SIDD.py --src_dir datasets/denoise/SIDD_Medium_Srgb/Data --tar_dir datasets/denoise/train
```

最后, 准备训练和验证所需json文件, 生成的json文件将保存在 `datasets/denoise/` 路径下.
```bash
python data/sidd/gen_json_sidd.py --split train
python data/sidd/gen_json_sidd.py --split val
```


### Low-Light Image Enhancement

首先, 下载 LOL 数据集 [google drive](https://drive.google.com/file/d/157bjO1_cFuSd0HWDUuAmcHRJDVyWpOxB/view), 将下载的数据集存放到 `$Painter_ROOT/datasets/light_enhance/` 路径下. 完成后的 LOL 文件结构如下所示:

```
light_enhance/
    our485/
        low/
        high/
    eval15/
        low/
        high/
```

Next, prepare json files for training and evaluation. The generated json files will be saved at `$Painter_ROOTdatasets/light_enhance/`.
```
python data/lol/gen_json_lol.py --split train
python data/lol/gen_json_lol.py --split val
```

数据集的目录结构如下：

```
├── nyu_depth_v2/
│   ├── sync/
│   ├── official_splits/
│   ├── nyu_depth_v2_labeled.mat
│   ├── nyuv2_sync_image_depth.json  # generated
│   ├── nyuv2_test_image_depth.json  # generated
├── ade20k/
│   ├── images/
│   ├── annotations/
│   ├── annotations_detectron2/  # generated
│   ├── annotations_with_color/  # generated
│   ├── ade20k_training_image_semantic.json  # generated
│   ├── ade20k_validation_image_semantic.json  # generated
├── ADEChallengeData2016/  # sim-link to $Painter_ROOT/datasets/ade20k
├── coco/
│   ├── train2017/
│   ├── val2017/
│   ├── annotations/
│       ├── instances_train2017.json
│       ├── instances_val2017.json
│       ├── person_keypoints_val2017.json
│       ├── panoptic_train2017.json
│       ├── panoptic_val2017.json
│       ├── panoptic_train2017/
│       ├── panoptic_val2017/
│   ├── panoptic_semseg_val2017/  # generated
│   ├── panoptic_val2017/  # sim-link to $Painter_ROOT/datasets/coco/annotations/panoptic_val2017
│   ├── pano_sem_seg/  # generated
│       ├── panoptic_segm_train2017_with_color
│       ├── panoptic_segm_val2017_with_color
│       ├── coco_train2017_image_panoptic_sem_seg.json
│       ├── coco_val2017_image_panoptic_sem_seg.json
│   ├── pano_ca_inst/  # generated
│       ├── train_aug0/
│       ├── train_aug1/
│       ├── ...
│       ├── train_aug29/
│       ├── train_org/
│       ├── train_flip/
│       ├── val_org/
│       ├── coco_train_image_panoptic_inst.json
│       ├── coco_val_image_panoptic_inst.json
├── coco_pose/
│   ├── person_detection_results/
│       ├── COCO_val2017_detections_AP_H_56_person.json
│   ├── data_pair/  # generated
│       ├── train_256x192_aug0/
│       ├── train_256x192_aug1/
│       ├── ...
│       ├── train_256x192_aug19/
│       ├── val_256x192/
│       ├── test_256x192/
│       ├── test_256x192_flip/
│   ├── coco_pose_256x192_train.json  # generated
│   ├── coco_pose_256x192_val.json  # generated
├── derain/
│   ├── train/
│       ├── input/
│       ├── target/
│   ├── test/
│       ├── Rain100H/
│       ├── Rain100L/
│       ├── Test100/
│       ├── Test1200/
│       ├── Test2800/
│   ├── derain_train.json
│   ├── derain_test_rain100h.json
├── denoise/
│   ├── SIDD_Medium_Srgb/
│   ├── train/
│   ├── val/
│   ├── denoise_ssid_train.json  # generated
│   ├── denoise_ssid_val.json  # generated
├── light_enhance/
│   ├── our485/
│       ├── low/
│       ├── high/
│   ├── eval15/
│       ├── low/
│       ├── high/
│   ├── enhance_lol_train.json  # generated
│   ├── enhance_lol_val.json  # generated
```

## 训练
下载预训练模型 [MAE ViT-Large model ](https://dl.fbaipublicfiles.com/mae/pretrain/mae_pretrain_vit_large.pth), 修改`$Painter_ROOT/train_painter_vit_large.sh`中finetune参数地址.
### 单机多卡
#### 普通训练

```
bash train_painter_vit_large.sh
```

#### 分布式训练
```
bash train_multi.sh
```

## 推理
下载推理模型[🤗 Hugging Face Models](https://huggingface.co/BAAI/Painter/blob/main/painter_vit_large.pth). The results on various tasks are summarized below:

## NYU Depth V2

To evaluate Painter on NYU Depth V2, you may first update the `$JOB_NAME` in `$Painter_ROOT/eval/nyuv2_depth/eval.sh`, then run:
```bash
bash eval/nyuv2_depth/eval.sh
```

## ADE20k Semantic Segmentation

To evaluate Painter on ADE20k semantic segmentation, you may first update the `$JOB_NAME` in `$Painter_ROOT/eval/ade20k_semantic/eval.sh`, then run:
```bash
bash eval/ade20k_semantic/eval.sh
```

## COCO Panoptic Segmentation

To evaluate Painter on COCO panoptic segmentation, you may first update the `$JOB_NAME` in `$Painter_ROOT/eval/coco_panoptic/eval.sh`, then run:
```bash
bash eval/coco_panoptic/eval.sh
```


## COCO Human Pose Estimation

为了评估Painter对COCO姿态的估计, 首先生成绘制的图像:
```bash
python -m torch.distributed.launch --nproc_per_node=8 --master_port=29500 --use_env eval/mmpose_custom/painter_inference_pose.py --ckpt_path models/painter_vit_large/painter_vit_large.pth
python -m torch.distributed.launch --nproc_per_node=8 --master_port=29500 --use_env eval/mmpose_custom/painter_inference_pose.py --ckpt_path models/painter_vit_large/painter_vit_large.pth --flip_test
```

Then, you may update the `job_name` and `ckpt_file` in `$Painter_ROOT/eval/mmpose_custom/configs/coco_256x192_test_offline.py`, and run:
```bash
cd $Painter_ROOT/eval/mmpose_custom
./tools/dist_test.sh configs/coco_256x192_test_offline.py none 1 --eval mAP
```

## Low-level Vision Tasks

### Deraining

To evaluate Painter on deraining, first generate the derained images.
```bash
python eval/derain/painter_inference_derain.py --ckpt_path models/painter_vit_large/painter_vit_large.pth
```

Then, update the path to derained images and ground truth in `$Painter_ROOT/eval/derain/evaluate_PSNR_SSIM.m` and run the following script in MATLAB.
```bash
$Painter_ROOT/eval/derain/evaluate_PSNR_SSIM.m
```


### Denoising

To evaluate Painter on SIDD denoising, first generate the denoised images.
```bash
python eval/sidd/painter_inference_sidd.py --ckpt_path models/painter_vit_large/painter_vit_large.pth
```

Then, update the path to denoising output and ground truth in `$Painter_ROOT/eval/sidd/eval_sidd.m` and run the following script in MATLAB.
```bash
$Painter_ROOT/eval/sidd/eval_sidd.m
```


### Low-Light Image Enhancement

To evaluate Painter on LoL image enhancement:
```bash
python eval/lol/painter_inference_lol.py --ckpt_path models/painter_vit_large/painter_vit_large.pth
```

#### 单卡推理

```
bash test.sh
```

## result

本地测试集测试结果单张展示：

<div align=center>
    <img src="./doc/origin.png"/>
</div>

<div align=center>
    <img src="./doc/results.png"/>
</div>

### 精度

基于项目提供的测试数据, 得到单卡测试结果如下：

|       | PSNR | SSIM | LPIPS  |
| :------: | :------: | :------: | :------: |
| ours | 29.04 | 0.7615 | 0.1294  |
| paper | 30.13 | 0.7981 | 0.1031 |


## 应用场景
### 算法类别
图像降噪

### 热点应用行业
交通,公安,制造

## 源码仓库及问题反馈
http://developer.hpccube.com/codes/modelzoo/maskeddenoising_pytorch.git

## 参考资料
https://github.com/haoyuc/MaskedDenoising.git