livepeer / livepeer/verification-classifier
Error rate analysis for BRISQUE model
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 8
- Forks
- 6
- PR merge metrics
- No merged PRs in 30d
Description
The current BRISQUE model created in #86 seems to have a higher error rate for synthetic videos that contain very abstract images vs. more "natural videos" that contain images from the physical world. But, it is unclear what constitutes a synthetic video with abstract images. For example, would video game replays fall into that category? One guess is that certain types of video games such as Pac-Man might, but other types of video games such as League of Legends might now. We should do some more analysis to narrow down the videos that have higher error rates.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the BRISQUE model created in issue #86 and the existing error-rate analysis. Compare results for abstract synthetic videos, natural videos, and possible video-game categories, then define which groups show higher error rates and document the findings.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- jupyter-notebook
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100