### Install Other HAWP Dependencies

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Install the remaining project dependencies listed in the requirement.txt file.

```bash
# Install other dependencies
pip install -r requirement.txt
```

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

### Install PyTorch with CUDA Support

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Install PyTorch version 1.12.0 with CUDA 11.6 support. Ensure your system's CUDA version matches the installation. This command specifies the exact versions for torch, torchvision, and torchaudio.

```bash
# Install pytorch, please be careful for the version of CUDA on your machine
pip install torch==1.12.0+cu116 torchvision==0.13.0+cu116 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu116
```

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

### Create and Activate Conda Environment for HAWP

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Set up a new Conda environment named 'hawp' with Python 3.9, activate it, and install the HAWP package in editable mode.

```bash
conda create -n hawp python==3.9
conda activate hawp
pip install -e .
```

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

### Verify PyTorch CUDA Availability

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Check if the installed PyTorch version correctly recognizes and can utilize the CUDA-enabled GPU on your machine.

```python
python -c "import torch; print(torch.cuda.is_available())" # Check if the installed pytorch supports CUDA.
```

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

### HAWPv3 Training Script Arguments

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.md

This displays the available command-line arguments for the HAWPv3 training script. It covers configuration file paths, experiment naming, pretrained model loading, and various training parameters like batch size, learning rate, and epochs.

```bash
python -m hawp.ssl.train --help
usage: train.py [-h] --datacfg DATACFG --modelcfg MODELCFG --name NAME
                [--pretrained PRETRAINED] [--overwrite] [--tf32]
                [--dtm {True,False}] [--batch-size BATCH_SIZE]
                [--num-workers NUM_WORKERS] [--base-lr BASE_LR]
                [--steps STEPS [STEPS ...]] [--gamma GAMMA] [--epochs EPOCHS]
                [--seed SEED] [--iterations ITERATIONS]

optional arguments:
  -h, --help            show this help message and exit
  --datacfg DATACFG     filepath of the data config
  --modelcfg MODELCFG   filepath of the model config
  --name NAME           the name of experiment
  --pretrained PRETRAINED
                        the pretrained model
  --overwrite           [Caution!] the option to overwrite an existed
                        experiment
  --tf32                toggle on the TF32 of pytorch
  --dtm {True,False}    toggle the deterministic option of CUDNN. This option
                        will affect the replication of experiments

training recipe:
  --batch-size BATCH_SIZE
                        the batch size of training
  --num-workers NUM_WORKERS
                        the number of workers for training
  --base-lr BASE_LR     the initial learning rate
  --steps STEPS [STEPS ...]
                        the steps of the scheduler
  --gamma GAMMA         the lr decay factor
  --epochs EPOCHS       the number of epochs for training
  --seed SEED           the random seed for training
  --iterations ITERATIONS
                        the number of training iterations

```

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

### Download Officially-Trained HAWP Checkpoints

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Execute the downloads.sh script to obtain the pre-trained model checkpoints for HAWPv2 and HAWPv3.

```bash
sh downloads.sh
```

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

### Build HAWP Docker Image

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Build a Docker image tagged as 'hawp:latest' from the provided Dockerfile. This command streams the Dockerfile content to the docker build command.

```bash
sudo docker build - < Dockerfile --tag hawp:latest
```

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

### Synthetic Data Training for HAWPv3

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.train.md

Initiates the synthetic training phase for HAWPv3. Configure dataset, model, learning rate, epochs, batch size, and training name.

```bash
python -m hawp.ssl.train \
    --datacfg hawp/ssl/config/synthetic_dataset.yaml \
    --modelcfg hawp/ssl/config/hawpv3.yaml \
    --base-lr 0.0004 \
    --epochs 10  \
    --batch-size 6 \
    --name hawpv3-round0
```

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

### Training HAWPv3 on Real-World Images

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.train.md

Trains HAWPv3 using pseudo wireframe labels generated from real-world images. Specify the generated data configuration, model configuration, learning rate, epochs, training name, and batch size.

```bash
python -m hawp.ssl.train  --datacfg data-ssl/export_datasets/{name}/{hash}-model-00010.yaml --modelcfg hawp/ssl/config/hawpv3.yaml --base-lr 0.0004 --epochs 30  --name hawpv3-round1 --batch-size 6
```

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

### Inference with HAWPv3 (Wireframe Dataset)

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.md

Use this command to perform inference and extract wireframes using the HAWPv3 model trained on the Wireframe dataset. Specify the checkpoint path, a confidence threshold, and the input image filename.

```bash
python -m hawp.ssl.predict --ckpt checkpoints/hawpv3-fdc5487a.pth \
            --threshold 0.05 \
            --img {filename.png}
```

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

### Download HAWPv2 Data using gdown

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Use the gdown tool to download the HAWPv2 training and testing datasets from Google Drive. Unzip the downloaded file to access the data.

```bash
gdown 134L-u9pgGtnzw0auPv8ykHqMjjZ2claO
unzip data.zip
```

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

### Batch Inference on DTU-24 Images

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.md

This command demonstrates running inference on multiple images from the DTU-24 dataset. It specifies the checkpoint, threshold, input image path pattern, output directory, and image format.

```bash
python -m hawp.ssl.predict --ckpt checkpoints/hawpv3-imagenet-03a84.pth  \
    --threshold 0.05  \
    --img ~/datasets/DTU/scan24/image/*.png \
    --saveto docs/figures/dtu-24 --ext png \
  
```

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

### Inference with HAWPv3 (ImageNet Dataset)

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.md

Use this command to perform inference and extract wireframes using the HAWPv3 model trained on ImageNet. Specify the checkpoint path, a confidence threshold, and the input image filename.

```bash
python -m hawp.ssl.predict --ckpt checkpoints/hawpv3-imagenet-03a84.pth \
            --threshold 0.05 \
            --img {filename.png}
```

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

### Homographic Adaptation (Batch Mode)

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.train.md

Performs homographic adaptation for pseudo wireframe generation in batch mode. Specify meta-architecture, dataset config, working directory, epoch, model config, minimum score threshold, and batch size.

```bash
python -m hawp.ssl.homoadp-bm --metarch HAWP-heatmap \
    --datacfg hawp/ssl/config/exports/wireframe-10iters.yaml \
    --workdir exp-ssl/hawpv3-round0 \
    --epoch 10 \
    --modelcfg exp-ssl/hawpv3-round0/model.yaml \
    --min-score 0.75   --batch-size=16
```

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

### Homographic Adaptation (Single Image Mode)

Source: https://github.com/cherubicxn/hawp/blob/main/docs/HAWPv3.train.md

Performs homographic adaptation for pseudo wireframe generation in single-image mode to minimize GPU memory. Specify meta-architecture, dataset config, working directory, epoch, model config, and minimum score threshold.

```bash
python -m hawp.ssl.homoadp --metarch HAWP-heatmap \
    --datacfg hawp/ssl/config/export/wireframe-10iters.yaml \
    --workdir exp-ssl/hawpv3-round0 \
    --epoch 10 \
    --modelcfg exp-ssl/hawpv3-round0/model.yaml \
    --min_score 0.75 
```

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

### HAWP Journal Citation

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Citation for the journal publication of the Holistically-Attracted Wireframe Parsing (HAWP) work. Use this when referencing the PAMI journal publication.

```bibtex
@article{HAWP-journal,
title = "Holistically-Attracted Wireframe Parsing: From Supervised to Self-Supervised Learning",
author = "Nan Xue and Tianfu Wu and Song Bai and Fu-Dong Wang and Gui-Song Xia and Liangpei Zhang and Philip H.S. Torr"
journal = "IEEE Trans. on Pattern Analysis and Machine Intelligence (PAMI)",
year = {2023}
}
```

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

### HAWP Conference Citation

Source: https://github.com/cherubicxn/hawp/blob/main/readme.md

Citation for the conference publication of the Holistically-Attracted Wireframe Parsing (HAWP) work. Use this when referencing the CVPR conference publication.

```bibtex
@inproceedings{HAWP,
title = "Holistically-Attracted Wireframe Parsing",
author = "Nan Xue and Tianfu Wu and Song Bai and Fu-Dong Wang and Gui-Song Xia and Liangpei Zhang and Philip H.S. Torr"
booktitle = "IEEE Conference on Computer Vision and Pattern Recognition (CVPR)",
year = {2020},
}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.