Perceptual Crack Detection for Rendered 3D Textured Meshes

Software release of the "Perceptual Crack Detection for Rendered 3D Textured Meshes" QoMEX 2024 paper.

Documentation

Please refer to our published paper, where we thoroughly explain the proposed PCD method.

Installation

cd path/to/this/repo
python3 -m venv venv
source venv/bin/activate
pip3 install -r requirements.txt

Usage

To compute the raw quality score, the new quality score, and to generate a video highlighting the crack artifacts, you can run the following command with the following arguments:

required arguments:

• --ref_add reference video path, string,
• --tst_add test video path, string.

optional arguments:

• -h or --help: shows the help message and exits,
• --method: initial quality assessment method to be enhanced by the weight map, string, options: lumaPSNR, SSIM, IW-SSIM! (default: IW-SSIM),
• --skip_fr: number of consecutive frames to skip in each interval when generating the outputs, int, (default: 20),
• --out_fps: desired frame rate (fps) for the output video, ignored when --no_vid flag is up, float, (default: 5.0),
• --no_vid: flag to disable generating the output video,
• --no_rqs: flag to disable computing the raw quality score.

Please note that the height and width of the output video frames are two times the height and width of the input video frames.

The following Example operates on one out of every 5 consecutive frames of the reference and test videos to compute the raw and enhanced SSIM quality score and to generate a video with 5.0 frame rate that highlights the crack artifacts:

cd path_to_this_repo
python3 main.py \
--ref_add data/statue_Ref.mp4 \
--tst_add data/statue_simpL6_qp8_qt6_decompJPEG_2048x2048_Q90.mp4 \
--skip_fr 5 \
--method SSIM 

Output of the preceding example:

Raw quality score: 0.9059885119062726

New quality score: 0.11615714537909715

A sample frame of the generated video:

Citation

If you use any part of this code or data in your research, please cite our publication.

Aknowledgement

The code for IW-SSIM and SSIM were adapted from the following references, respectively:

[1] PyTorch Image Quality: Metrics for Image Quality Assessment. Kastryulin, Sergey and Zakirov, Jamil and Prokopenko, Denis and Dylov, Dmitry V. arXiv 2022. https://arxiv.org/abs/2208.14818.

[2] https://github.com/mubeta06/python/blob/master/signal_processing/sp/ssim.py.