This repo contains the code to train and test the frame-level NDNetGaming. The code for training is placed in the training folder. Please report any problem you faced or suggestion you have to improve the code.

In order to run the code there are two options:

• Test the model on a video, which only the model (path together with name), video path, video name and test type (select video) must be specified. To do so, you can run it as follow:

    python test.py 
        --model=./models/subjectiveDemo2_DMOS_Final.model
        --videopath=./videos/ 
        --videoname=sample1.mp4 
        --test_type=video

• Test the model based on the extracted frames in nested folders, which requires specifying the model (path together with name), path of the folder that contains folders of frames, image format (e.g. png) and test type (image_folders). -- It is very important to place all folders contain extracted frames in a specific folder. Please follow the example provided in this repo.

    python test.py 
        --model=./models/subjectiveDemo2_DMOS_Final.model
        --folder=./frames/ 
        --imageformat=png 
        --test_type=image_folders

For more help run:

   python test.py -h

The model gives you frame level quality and save it in a csv file. The NDNetGaming follows the MOS score scale, ranging from 1 to 5.

• If you run the model for a video (using test_tyep = video), you may get memory error if the video is too lenghty or large. We recommend extracting frames beforehand and use the image-folders mode for testing.
• Please note that the model only predicts the frame quality. The temporal pooling component is only available if --test_type is set to video.
• The code is made for Windows users and it needs revision to be used for Linux users.

Install python and pip, if they are not already installed. Follow the platform specific installation instructions. The following step should be performed to prepare the setup.

    git clone https://github.com/stootaghaj/NDNetGaming.git 
    pip install -r requirements.txt

Please cite the paper below if you use the code or to get more insight about the model:

	@inproceedings{NDNetgaming,
		title={{NDNetGaming - Development of a No-Reference Deep CNN for Gaming Video Quality Prediction}},
		author={Utke, Markus and Zadtootaghaj, Saman and Schmidt, Steven and Bosse, Sebastian and Moeller, Sebastian  },
		booktitle={Multimedia Tools and Applications},
		year={2020},
		organization={Springer},
	  }

MIT License

Copyright 2020 (c) Saman Zadtootaghaj.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

• Markus Utke
• Saman Zadtootaghaj