Comments (11)
I've already emailed @aizvorski about this. He hasn't responded.
from scikit-video.
Thanks for your quick reply.
By the way, I am a master candicate from China.
I got your are a PHD on I&VQA just now. Great GPA.
I did some research about IQA for a short time.
And I am doing something about video question answer recently. And I have a subtask to confirm the video coherent, a video may be made up many small videos. I use SSIM
and a threshold to confirm the video is from the same small video or not. It's not good enough.
Do you have any idea to better solve this?
Thanks so much.
from scikit-video.
I'm not sure exactly what you mean. A video may be made up of many small videos? Do you mean videos with different scenes, or do you mean finding a video clip inside the original video?
from scikit-video.
The subtask is more similar to the latter description.
Due to the dirty dataset, a video may be made up of many sub videos, which are different.
For example:
For something to do about films, the video in the dataset may have this video:
many clips from different films make up a new video. And the new video is what I have.
I have to pick up these different film clips from the video. SSIM is not good enough.
from scikit-video.
Well, if you already have the clip that you interested in finding within the new video, you can just use mean squared error (MSE) or sum of absolute differences (SAD). In other words, the clip is your template, and you want to correlate it against the new video. The correlation will be highest when mean squared error is lowest.
from scikit-video.
I have tried MSE
, but not SAD
. I will have a try to know whether useful.
The MSE
result is really rough.
I don't know how many images are from the same clip, and I don't have clips. I only have the new video.
Images may have low mse value, and they are from different clips.
SSIM has a better result, but not good enough.
from scikit-video.
It sounds like I completely misunderstood your problem. You can try a scene detector to split the clips in the new video. There is one already in sk-video.
from scikit-video.
@beyondmetis I added you to the scikit-video pypi as maintainer so you can go ahead and change the info over to point here and upload the latest release. I'll open a separate issue for a few cleanup things that we'll have to do around that - I think not blocking.
from scikit-video.
I saw the api. I'll have a try for this.
Actually, before SSIM, I use the key-frame
to roughly detect small clips.
And then use SSIM to merge this rough clips. Obviously, SSIM is not enough to clip.
from scikit-video.
@alex-at-thimble Thanks! I'll need to make a few updates to documentation as well, which should not be a big deal.
from scikit-video.
I've pushed the current version of sk-video into pypi as scikit-video. @oneTaken if you'd like to take this unrelated conversation offline, you can always email me directly.
from scikit-video.
Related Issues (20)
- output dict params
- measure/data is missing after installation from Git HOT 1
- Numpy warnings due to deprecated aliases HOT 4
- scipy version compatibility issue HOT 2
- How to encode video with FFmpegWriter so it is compatible with powerpoint 2016 HOT 1
- Typo in documentation: skvideo.io.write
- ffmpeg/ffprobe not found in path HOT 1
- Document inputdict and outputdict of FFmpegWriter HOT 1
- The extracted frames changed the colour of the video
- set fps in vwrite
- ValueError: No way to determine width or height from video. Need `-s` in `inputdict`. Consult documentation on I/O. HOT 2
- module 'skvideo' has no attribute 'io' HOT 1
- skvideo.io.ffprobe does not properly capture multiple streams; captures only the last stream of a '@codec_type'
- Load specific index (frame number) range HOT 5
- DeprecationWarning: tostring() is deprecated. Use tobytes() instead. HOT 1
- [Question] Add multiple metadata kv to file
- issue with reading video with variable frame rate
- FFmpegReader & FFmpegWriter no audio stream;
- Broken pipe errors in reader on enabled verbosity
- no sound in video output
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from scikit-video.