TransNet V2 Visualization for Shot Boundary Detection

---

Introduction: A simple Python interface to show shot boundary detection based on TransNet V2.
Repo size: 55 MB
Hint: We just use the pretrained model here. If you want to train it, please see the TransNet V2 repo.
Modifier: Shallwe
Author: Tomáš Souček
Paper: TransNet V2: An effective deep network architecture for fast shot transition detection

---

Environment

Recommended to run in Linux. Tested in Ubuntu 18.04 .

Necessary environments include TensorFlow, ffmpeg, OpenCV, PIL, Tkinter and so on, run these commands to install them.

pip install tensorflow==2.1
pip install opencv-python ffmpeg-python pillow
sudo apt install ffmpeg python-tk

How to run

# Run it in the project directory
python3 main.py

Files Prepared:

• Original video. (MP4 format recommended)

• Label files of SBD. (Same directory as video. Named as "VIDEO_NAME+.truth.txt")

In directory test_data/ , I provide a video bbc_02_clip.mp4 as an example (Clipped from BBC Planet Earth Test Dataset. You can download more test videos here.) to help you test this program.

Process:

• Input location of video file. (Need to provide relative path from current work directory such as test_data/bbc_02_clip.mp4 )

• Select "Device Select" (Recommend default)

• Click "Shot boundary detection"

• Wait for the evaluation of the model (until SBD success appears in cmd)

• Click "Visualize"

• You can click "Start" or "Pause" while video is playing

• Click "Quit"

Files created:

• VIDEO_PATH/video.mp4.scenes.txt file containing a list of scenes - pairs of start-frame-index, end-frame-index (indexed from zero, both limits inclusive).
• VIDEO_PATH/video.mp4.predictions.txt file with each line containing raw predictions for corresponding frame (fist number is from the first 'single-frame-per-transition' head, the second from 'all-frames-per-transition' head)
• optionally it creates visualization in file /path/to/video.mp4.vis.png

Credits

If found useful, please cite the author.

• GitHub Website: soCzech/TransNetV2

• This paper: TransNet V2: An effective deep network architecture for fast shot transition detection

  @article{soucek2020transnetv2,
      title={TransNet V2: An effective deep network architecture for fast shot transition detection},
      author={Sou{\v{c}}ek, Tom{\'a}{\v{s}} and Loko{\v{c}}, Jakub},
      year={2020},
      journal={arXiv preprint arXiv:2008.04838},
  }
  

• ACM Multimedia paper of the older version: A Framework for Effective Known-item Search in Video

• The older version paper: TransNet: A deep network for fast detection of common shot transitions