### Setup AniLines Environment

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/AniLines/README.md

Clone the repository, create a conda environment, and install dependencies. Ensure you have Python 3.12 or compatible.

```bash
git clone https://github.com/zhenglinpan/AniLines-Anime-Line-Extractor.git
cd AniLines-Anime-Line-Extractor

conda create -n anilines python=3.12 -y
conda activate anilines

pip install -r requirements.txt
```

--------------------------------

### Install Detectron2 from Source (Git Clone)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install Detectron2 by cloning the repository and installing it in editable mode. This is useful for development or when needing the latest features.

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

--------------------------------

### Install PanopticAPI

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/docs/tutorials/builtin_datasets.md

Install the panopticapi library from its GitHub repository. This is required for setting up datasets for PanopticFPN.

```bash
pip install git+https://github.com/cocodataset/panopticapi.git
```

--------------------------------

### Install Detectron2 from Source (GitHub)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/INSTALL.md

Installs Detectron2 directly from its GitHub repository. Use '--user' if you lack sufficient permissions.

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

--------------------------------

### Install TensorMask Project

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/TensorMask/README.md

Install the TensorMask project within Detectron2. Ensure Detectron2 is installed and the dataset is set up before compiling the TensorMask-specific op.

```bash
pip install -e /path/to/detectron2/projects/TensorMask
```

--------------------------------

### Install Detectron2 from Source (Pip)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install Detectron2 directly from its GitHub repository using pip. Use the --user flag if you lack system-wide permissions.

```bash
python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
# (add --user if you don't have permission)
```

--------------------------------

### Install Detectron2 (CPU, PyTorch 1.8)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel for CPU-only environments with PyTorch 1.8. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cpu/torch1.8/index.html
```

--------------------------------

### Install Detectron2 (CPU, PyTorch 1.10)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel for CPU-only environments with PyTorch 1.10. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cpu/torch1.10/index.html
```

--------------------------------

### Install LVIS API

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/docs/tutorials/builtin_datasets.md

Install the lvis-api library from its GitHub repository. This is needed for working with LVIS datasets in Detectron2.

```bash
pip install git+https://github.com/lvis-dataset/lvis-api.git
```

--------------------------------

### Install Detectron2 (CPU, PyTorch 1.9)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel for CPU-only environments with PyTorch 1.9. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cpu/torch1.9/index.html
```

--------------------------------

### Training Schedule Example

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docs/notes/benchmarks.md

This example demonstrates a multi-stage training schedule. Adjust the learning rate, epochs, layers to fine-tune, and augmentation as needed for your specific task.

```python
model.train(dataset_train, dataset_val,
            learning_rate=config.LEARNING_RATE,
            epochs=40,
            layers='3+',
            augmentation=augmentation)
```

--------------------------------

### Install Cityscapes Scripts

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/datasets/README.md

Install the necessary scripts for processing the Cityscapes dataset.

```bash
pip install git+https://github.com/mcordts/cityscapesScripts.git
```

--------------------------------

### Install Detectron2 from Local Clone (macOS)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/INSTALL.md

Specific installation command for macOS users when installing from a local clone.

```bash
# CC=clang CXX=clang++ python -m pip install -e .
```

--------------------------------

### Install Detectron2 (CUDA 10.1, PyTorch 1.8)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 10.1 and PyTorch 1.8. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/torch1.8/index.html
```

--------------------------------

### Install and Prepare Panoptic FPN Annotations

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/datasets/README.md

Install the panopticapi library and run a script to extract semantic annotations from COCO panoptic annotations for PanopticFPN.

```bash
pip install git+https://github.com/cocodataset/panopticapi.git
```

```python
python datasets/prepare_panoptic_fpn.py
```

--------------------------------

### Command Line Config Overwrite Example

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docs/tutorials/configs.md

Illustrates how to overwrite configuration values using command-line arguments with demo.py.

```bash
./demo.py --config-file config.yaml [--other-options] \
  --opts MODEL.WEIGHTS /path/to/weights INPUT.MIN_SIZE_TEST 1000
```

--------------------------------

### Trainer Log Output Example

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/DensePose/doc/BOOTSTRAPPING_PIPELINE.md

Example log output from the DensePose trainer showing the proportion of data from different sources in a batch.

```text
[... densepose.engine.trainer]: batch/ 1.8, batch/base_coco_2017_train 6.4, batch/densepose_coco_2014_train 3.85

```

--------------------------------

### Install DensePose as a Python Package

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/DensePose/doc/GETTING_STARTED.md

Install DensePose from its GitHub repository using pip. This command installs the package and makes it importable as 'densepose'.

```bash
pip install git+https://github.com/facebookresearch/detectron2@main#subdirectory=projects/DensePose
```

--------------------------------

### Install Detectron2 (CUDA 10.2, PyTorch 1.10)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 10.2 and PyTorch 1.10. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu102/torch1.10/index.html
```

--------------------------------

### Install Detectron2 (CUDA 10.2, PyTorch 1.8)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 10.2 and PyTorch 1.8. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu102/torch1.8/index.html
```

--------------------------------

### Manually Install Flash Attention

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/README.md

If encountering errors during Flash Attention installation, manually download the appropriate wheel file based on your Python, CUDA, and Torch versions, then install using pip.

```shell
pip install flash_attn-2.7.3+cu12torch2.2cxx11abiFALSE-cp312-cp312-linux_x86_64.whl
```

--------------------------------

### Install Detectron2 (CUDA 11.1, PyTorch 1.8)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 11.1 and PyTorch 1.8. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.8/index.html
```

--------------------------------

### Install Detectron2 (CUDA 11.1, PyTorch 1.9)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 11.1 and PyTorch 1.9. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.9/index.html
```

--------------------------------

### Install Detectron2 (CUDA 11.1, PyTorch 1.10)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 11.1 and PyTorch 1.10. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.10/index.html
```

--------------------------------

### Train TridentNet with ResNet-50

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/TridentNet/README.md

Example command to launch end-to-end TridentNet training using ResNet-50 backbone on 8 GPUs. Ensure the config file path is correct.

```bash
python /path/to/detectron2/projects/TridentNet/train_net.py --config-file configs/tridentnet_fast_R_50_C4_1x.yaml --num-gpus 8
```

--------------------------------

### Install Pre-Built Detectron2 (Linux CUDA 10.1)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/INSTALL.md

Installs a pre-built Detectron2 wheel for Linux systems with CUDA 10.1. Replace 'cu101' with other CUDA versions or 'cpu' as needed.

```bash
# for CUDA 10.1:
python -m pip install detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/index.html
```

--------------------------------

### Install Detectron2 (CUDA 10.2, PyTorch 1.9)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 10.2 and PyTorch 1.9. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu102/torch1.9/index.html
```

--------------------------------

### Compile and Run Caffe2 Mask R-CNN Example

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/docs/tutorials/deployment.md

Compiles and runs a C++ example for CPU/GPU inference using a converted Caffe2 Mask R-CNN model. Ensure all dependencies and build configurations are met.

```bash
sudo apt update && sudo apt install libgflags-dev libgoogle-glog-dev libopencv-dev
pip install mkl-include
wget https://github.com/protocolbuffers/protobuf/releases/download/v3.6.1/protobuf-cpp-3.6.1.tar.gz
tar xf protobuf-cpp-3.6.1.tar.gz
export CPATH=$(readlink -f ./protobuf-3.6.1/src/):$HOME/.local/include
export CMAKE_PREFIX_PATH=$HOME/.local/lib/python3.6/site-packages/torch/
mkdir build && cd build
cmake -DTORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST .. && make

# To run:
./caffe2_mask_rcnn --predict_net=./model.pb --init_net=./model_init.pb --input=input.jpg
```

--------------------------------

### Install Detectron2 (CUDA 11.3, PyTorch 1.10)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/INSTALL.md

Install a pre-built Detectron2 wheel compatible with CUDA 11.3 and PyTorch 1.10. This is for Linux systems only.

```bash
python -m pip install detectron2 -f 
  https://dl.fbaipublicfiles.com/detectron2/wheels/cu113/torch1.10/index.html
```

--------------------------------

### Build and Launch Detectron2 Deployment Docker Container

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docker/README.md

Builds a Docker image for deploying Detectron2 with C++ examples and launches an interactive container with GPU support.

```bash
# Build:
docker build -t detectron2-deploy:v0 -f deploy.Dockerfile .
# Launch:
docker run --gpus all -it detectron2-deploy:v0
```

--------------------------------

### Launch TensorMask BiPyramid Training

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/TensorMask/README.md

Example command to launch TensorMask BiPyramid training with a ResNet-50 backbone and a 1x schedule on 8 GPUs. Ensure the config file path is correct.

```bash
python /path/to/detectron2/projects/TensorMask/train_net.py --config-file configs/tensormask_R_50_FPN_1x.yaml --num-gpus 8
```

--------------------------------

### Install and Prepare Cocofied LVIS Annotations

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/datasets/README.md

Install the lvis-api and run a script to prepare 'cocofied' LVIS annotations for evaluating models trained on COCO.

```bash
pip install git+https://github.com/lvis-dataset/lvis-api.git
```

```python
python datasets/prepare_cocofied_lvis.py
```

--------------------------------

### Show Bounding Box and V Coordinate Visualization

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_APPLY_NET.md

Example command to visualize bounding boxes and the estimated V coordinates for body parts. Specify 'bbox,dp_v' for the visualizations.

```bash
python apply_net.py show configs/densepose_rcnn_R_50_FPN_s1x.yaml DensePose_ResNet50_FPN_s1x-e2e.pkl image.jpg bbox,dp_v -v
```

--------------------------------

### Show Bounding Box and U Coordinate Visualization

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_APPLY_NET.md

Example command to visualize bounding boxes and the estimated U coordinates for body parts. Use 'bbox,dp_u' for the visualizations argument.

```bash
python apply_net.py show configs/densepose_rcnn_R_50_FPN_s1x.yaml DensePose_ResNet50_FPN_s1x-e2e.pkl image.jpg bbox,dp_u -v
```

--------------------------------

### Visualize Bounding Box and Segmentation

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of visualizing bounding boxes and segmentation masks for a specific image. The -v flag enables verbose output.

```bash
python query_db.py show densepose_coco_2014_train image_id:int=322 bbox,dp_segm -v
```

--------------------------------

### Start AniLines Gradio UI

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/AniLines/README.md

Launch the Gradio web interface for AniLines locally. This allows for interactive processing of images, videos, and batch files.

```bash
python app.py
```

--------------------------------

### Visualize Bounding Box and Point Labels

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of visualizing bounding boxes and points colored by their containing part for a specific image. The -v flag enables verbose output.

```bash
python query_db.py show densepose_coco_2014_train image_id:int=322 bbox,dp_i -v
```

--------------------------------

### Show Bounding Box and Contour Visualization

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_APPLY_NET.md

Example command to visualize bounding boxes and contour plots with color-coded U and V coordinates. Use 'dp_contour,bbox' for the visualizations.

```bash
python apply_net.py show configs/densepose_rcnn_R_50_FPN_s1x.yaml DensePose_ResNet50_FPN_s1x-e2e.pkl image.jpg dp_contour,bbox -v
```

--------------------------------

### Visualize Bounding Box and U Coordinate Points

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of visualizing bounding boxes and points colored by their U coordinate for a specific image. The -v flag enables verbose output.

```bash
python query_db.py show densepose_coco_2014_train image_id:int=322 bbox,dp_u -v
```

--------------------------------

### Visualize Bounding Box and V Coordinate Points

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of visualizing bounding boxes and points colored by their V coordinate for a specific image. The -v flag enables verbose output.

```bash
python query_db.py show densepose_coco_2014_train image_id:int=322 bbox,dp_v -v
```

--------------------------------

### DensePose Bootstrapping Configuration

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/DensePose/doc/BOOTSTRAPPING_PIPELINE.md

Example YAML configuration for bootstrapping DensePose datasets. Specifies dataset details, image loading, inference parameters, data sampling, and filtering criteria.

```yaml
BOOTSTRAP_DATASETS:
  - DATASET: "chimpnsee"
    RATIO: 1.0
    IMAGE_LOADER:
      TYPE: "video_keyframe"
      SELECT:
        STRATEGY: "random_k"
        NUM_IMAGES: 4
      TRANSFORM:
        TYPE: "resize"
        MIN_SIZE: 800
        MAX_SIZE: 1333
      BATCH_SIZE: 8
      NUM_WORKERS: 1
    INFERENCE:
      INPUT_BATCH_SIZE: 1
      OUTPUT_BATCH_SIZE: 1
    DATA_SAMPLER:
      # supported types:
      #   densepose_uniform
      #   densepose_UV_confidence
      #   densepose_fine_segm_confidence
      #   densepose_coarse_segm_confidence
      TYPE: "densepose_uniform"
      COUNT_PER_CLASS: 8
    FILTER:
      TYPE: "detection_score"
      MIN_VALUE: 0.8
BOOTSTRAP_MODEL:
  WEIGHTS: https://dl.fbaipublicfiles.com/densepose/evolution/densepose_R_50_FPN_DL_WC1M_3x_Atop10P_CA/217578784/model_final_9fe1cc.pkl

```

--------------------------------

### Basic CfgNode Usage

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docs/tutorials/configs.md

Demonstrates obtaining default configurations, adding custom settings, merging from files and lists, and saving/printing configurations.

```python
from detectron2.config import get_cfg
cfg = get_cfg()    # obtain detectron2's default config
cfg.xxx = yyy      # add new configs for your own custom components
cfg.merge_from_file("my_cfg.yaml")   # load values from a file

cfg.merge_from_list(["MODEL.WEIGHTS", "weights.pth"])   # can also load values from a list of str
print(cfg.dump())  # print formatted configs
with open("output.yaml", "w") as f:
  f.write(cfg.dump())   # save config to file
```

--------------------------------

### Install pycocotools

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/INSTALL.md

Installs the pycocotools package, which is required for COCO dataset operations and visualizations.

```bash
pip install cython; pip install -U 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
```

--------------------------------

### Show Bounding Box and Segmentation

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_APPLY_NET.md

Example command to visualize bounding boxes and segmentation masks for detected persons. Requires a configuration file, model file, input image, and 'bbox,dp_segm' as visualizations.

```bash
python apply_net.py show configs/densepose_rcnn_R_50_FPN_s1x.yaml DensePose_ResNet50_FPN_s1x-e2e.pkl image.jpg bbox,dp_segm -v
```

--------------------------------

### Install New Dependencies in Dockerfile

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docker/README.md

Adds a RUN command to the Dockerfile to install new system dependencies using apt-get. This makes persistent changes to the image.

```dockerfile
RUN sudo apt-get update && sudo apt-get install -y vim
```

--------------------------------

### Install Project Requirements

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/README.md

Create a conda environment and install project dependencies using the provided requirements file or environment.yaml. Ensure Python, CUDA, and Torch versions are compatible.

```shell
conda create -n magictryon python==3.12.9
conda activate magictryon
pip install -r requirements.txt
```

```shell
conda env create -f environment.yaml
```

--------------------------------

### Example of BGR Input Format

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/docs/tutorials/extend.md

This snippet shows how to specify the input format as BGR when configuring Detectron2.

```python
input_format="BGR",
   )
```

--------------------------------

### Video Try-On Inference

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/README.md

Execute the video try-on demo using the specified Python scripts. Assign the CUDA device and adjust inference parameters in the script as needed.

```powershell
CUDA_VISIBLE_DEVICES=0 python inference/video_tryon/predict_video_tryon_up.py
```

```powershell
CUDA_VISIBLE_DEVICES=1 python inference/video_tryon/predict_video_tryon_low.py
```

--------------------------------

### Compile TensorMask Op

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/TensorMask/README.md

Compile the TensorMask-specific op 'swap_align2nat' after installing Detectron2 and setting up the dataset. This is a prerequisite for using TensorMask.

```bash
cd /path/to/detectron2/projects/TensorMask
python setup.py build develop
```

--------------------------------

### Build All Wheels and Generate Index

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/dev/packaging/README.md

Execute these scripts to build wheels for all supported CUDA and Python combinations and then generate an index file for them.

```bash
./dev/packaging/build_all_wheels.sh
./dev/packaging/gen_wheel_index.sh /path/to/wheels
```

--------------------------------

### Print First 10 Entries

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of printing at most 10 entries from the densepose_coco_2014_train dataset. The -v flag enables verbose output.

```bash
python query_db.py print densepose_coco_2014_train \* --max-entries 10 -v
```

--------------------------------

### Basic CfgNode Usage in Detectron2

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/docs/tutorials/configs.md

Demonstrates obtaining default configurations, adding custom settings, merging from files and lists, and printing the formatted configuration. Use this for initializing and modifying Detectron2's configuration settings.

```python
from detectron2.config import get_cfg
cfg = get_cfg()    # obtain detectron2's default config
cfg.xxx = yyy      # add new configs for your own custom components
cfg.merge_from_file("my_cfg.yaml")   # load values from a file

cfg.merge_from_list(["MODEL.WEIGHTS", "weights.pth"])   # can also load values from a list of str
print(cfg.dump())  # print formatted configs
```

--------------------------------

### Train Panoptic-DeepLab Model

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/projects/Panoptic-DeepLab/README.md

Use this command to train a Panoptic-DeepLab model on a dataset like Cityscapes. Ensure Detectron2 is installed and data is prepared.

```bash
cd /path/to/detectron2/projects/Panoptic-DeepLab
python train_net.py --config-file configs/Cityscapes-PanopticSegmentation/panoptic_deeplab_R_52_os16_mg124_poly_90k_bs32_crop_512_1024_dsconv.yaml --num-gpus 8
```

--------------------------------

### Build and Run Detectron2 Docker Container (docker ≥ 19.03)

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/docker/README.md

Builds the Detectron2 Docker image and runs it with GPU support. Grants access to the host X server for displaying images.

```bash
cd docker/
# Build:
docker build --build-arg USER_ID=$UID -t detectron2:v0 .
# Run:
docker run --gpus all -it \
	--shm-size=8gb --env="DISPLAY" --volume="/tmp/.X11-unix:/tmp/.X11-unix:rw" \
	--name=detectron2 detectron2:v0

# Grant docker access to host X server to show images
xhost +local:`docker inspect --format='{{ .Config.Hostname }}' detectron2`
```

--------------------------------

### Print Entries by File Name

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/gen_mask/preprocess/humanparsing/mhp_extension/detectron2/projects/DensePose/doc/TOOL_QUERY_DB.md

Example of printing all entries where the file_name matches a specific value. The -v flag enables verbose output.

```bash
python query_db.py print densepose_coco_2014_train file_name=COCO_train2014_000000000036.jpg -v
```

--------------------------------

### Run Inference Demo with Pre-trained Models

Source: https://github.com/vivocameraresearch/magic-tryon/blob/main/inference/customize/detectron2/GETTING_STARTED.md

Use the demo.py script to run inference on images using pre-trained models from the model zoo. Specify input files, configuration, and model weights. Common options include running on webcam, video, or CPU, and saving outputs.

```bash
cd demo/
python demo.py --config-file ../configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml \
  --input input1.jpg input2.jpg \
  [--other-options] \
  --opts MODEL.WEIGHTS detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl
```