Practical overview
When it fits
Use it when container FPS alone is not enough for examining the temporal scale implied by a video's motion, particularly in generated-video evaluation.
Who it is for
- Researchers studying video motion and physical frame-rate estimation.
- Video-generation teams assessing temporal consistency in generated clips.
- Developers comparing motion-derived frame-rate estimates across videos.
Documented uses
- Estimate a video's physical frame rate from its visual dynamics.
- Run predictions on one video or a directory and save directory results to CSV.
- Evaluate generated video outputs with the PhyFPS-Bench-Gen prompt set.
Requirements & limits
The repository is public. Its README describes a released checkpoint that downloads from Hugging Face on first inference, with a manual download option. The strongest internal model is not fully released; the README directs interested users to contact the project about access. No license information is provided in the repository evidence.
- The output is a visual-motion-derived frame-rate estimate, not a change to the video itself.
- The released checkpoint is intended for demo and leaderboard reproduction and differs from the strongest internal model.
- The benchmark workflow assumes users generate their own videos before running prediction on them.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install the inference requirements
Prepare the repository’s inference environment before using its documented prediction commands.
Before you start
- A Linux environment with Python available.
- Git and pip.
Steps
Clone https://github.com/taco-group/Visual_Chronometer.git.
Change into the Visual_Chronometer/inference directory.
Install the listed requirements with pip install -r requirements.txt.
Expected outcome
The repository’s documented inference requirements are installed; prediction commands are the next documented stage.
Guide limitations
- The repository says the released checkpoint is for demo and leaderboard reproduction and is not identical to its strongest internal model.
Estimate PhyFPS for a video directory
Produce physical frame-rate estimates for videos in a directory and save the results to CSV.
Before you start
- The inference environment and requirements described in the repository installation instructions.
- A directory of video files.
Steps
Change into the inference directory.
Run python predict.py --video_dir path/to/videos/ --output_csv results.csv, substituting the input directory and desired CSV path.
Expected outcome
The documented command requests predictions for the directory and writes results to the named CSV file.
Guide limitations
- The README does not specify supported video formats or further runtime requirements.
- The released checkpoint is provided for demo and leaderboard reproduction; it is not the strongest internal model.
Compare options for Model inference and serving
These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.
| Project | Documented purpose | Platform | Access and requirements |
|---|---|---|---|
| Pulse Of Motion | Estimate a video's physical frame rate from its visual dynamics. | Linux | The repository is public. Its README describes a released checkpoint that downloads from Hugging Face on first inference, with a manual download option. The strongest internal model is not fully released; the README directs interested users to contact the project about access. No license information is provided in the repository evidence. |
| Neural Image Assessment | Train the model using AVA images and rating annotations. | Cloud, Linux, macOS, Windows |
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 18 October 2026
- Original website title
- taco-group/Pulse-of-Motion · Official homepage title · retained in its source language