Comments (8)
erikalu
commented on June 9, 2024
Hi, what data are you training on? Is it the tennis demo or your custom video?
H_0 and H_1 are homographies for the current and next frame, respectively. This error is probably because the homography file isn't formatted correctly.
from omnimatte.
ai1361720220000
commented on June 9, 2024
Thanks for your reply. I used the tennis demo for training the network. Is there any wrong with the origin tennis data in this repos? Also, after I change the parameter max_dataset_size to 70, the code run without error.
BTW, I have another question. How many epochs does the network run based on the tennis demo?
发自我的iPhone
…------------------ Original ------------------
From: Erika Lu ***@***.***>
Date: Mon,Jul 26,2021 10:33 PM
To: erikalu/omnimatte ***@***.***>
Cc: Alxendx ***@***.***>, Author ***@***.***>
Subject: Re: [erikalu/omnimatte] IndexError: list index out of range (#3)
Hi, what data are you training on? Is it the tennis demo or your custom video?
H_0 and H_1 are homographies for the current and next frame, respectively. This error is probably because the homography file isn't formatted correctly.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
from omnimatte.
erikalu
commented on June 9, 2024
That's strange, I just checked and I don't get this error, even without changing max_dataset_size.
The default number of training steps is 12k (the number of epochs it corresponds to will depend on the batch size you use). You may want to increase the number of training steps if you're using a small batch size.
from omnimatte.
ai1361720220000
commented on June 9, 2024
Emmm it’s very confusing for the error. Also, 12K with batch size of 8 may cost more than 10 hours which is so expensive for a certain experiment.
发自我的iPhone
…------------------ Original ------------------
From: Erika Lu ***@***.***>
Date: Mon,Jul 26,2021 11:32 PM
To: erikalu/omnimatte ***@***.***>
Cc: Alxendx ***@***.***>, Author ***@***.***>
Subject: Re: [erikalu/omnimatte] IndexError: list index out of range (#3)
That's strange, I just checked and I don't get this error, even without changing max_dataset_size.
The default number of training steps is 12k (the number of epochs it corresponds to will depend on the batch size you use). You may want to increase the number of training steps if you're using a small batch size.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
from omnimatte.
erikalu
commented on June 9, 2024
What GPU are you using? It took about 2 hours to run on 2 P100 GPUs.
from omnimatte.
ai1361720220000
commented on June 9, 2024
I run on a Tesla V100
发自我的iPhone
…------------------ Original ------------------
From: Erika Lu ***@***.***>
Date: Mon,Jul 26,2021 11:49 PM
To: erikalu/omnimatte ***@***.***>
Cc: Alxendx ***@***.***>, Author ***@***.***>
Subject: Re: [erikalu/omnimatte] IndexError: list index out of range (#3)
What GPU are you using? It took about 2 hours to run on 2 P100 GPUs.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
from omnimatte.
ai1361720220000
commented on June 9, 2024
Hello, could you tell how to get zbar.pth for a new dataset. What's the zbar meaning? i also confused whether zbar would be update if it is set randomly in the network?
from omnimatte.
erikalu
commented on June 9, 2024
zbar.pth is randomly initialized and saved if it doesn't already exist, otherwise the existing file is loaded. These values aren't updated during training. It is random fixed noise that represents the input for the background layer.
from omnimatte.
Related Issues (20)
- 🔥 every video should train the new model? HOT 2
- Out of memory about train HOT 6
- Support Windows
- Optical flow HOT 2
- Computing homographies HOT 3
- Error for connecting server HOT 1
- Pretrained Model for Tennis Example
- how use Detectron2 output Binary Mask image HOT 1
- How to compute flow and save the forward/backward .flo files HOT 3
- ERROR: Could not find a version that satisfies the requirement opencv-python==4.5.1 HOT 1
- The video for computing homographies.txt should be the orignal one or the resized one? HOT 1
- Duration of the processed video is Longer than source video HOT 3
- Why every video should train a new model?
- Discrepancy between code and paper HOT 1
- The minimum memory of GPU HOT 1
- Can this work also on Windows 10 + Anaconda ? HOT 1
- Problem about H calculation
- Omnimatte without NVIDIA card HOT 1
- Open CV Can't be found HOT 2
- MaskRCNN + STM helper code for extracting frame-wise objects HOT 1
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 omnimatte.