关于插帧任务中损失函数的疑问
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1k
- Forks
- 131
- PR merge metrics
- No merged PRs in 30d
Description
作者你好，感谢你为社区所做的开源工作，关于在训练插帧模型的损失函数方面 有些问题想向你请教。
我目前采用的模型也是类似RIFE这种无预训练光流模块的插帧模型，实验发现：
1）用Laplacian Loss会使模型在小物体细节上建模得更好（比如滚动的字幕），但容易出现奇怪的光流错误；
2）用L1 Loss + Vgg Loss组合 则在插帧效果上更稳定，很大程度缓解了光流出错的概率，但在小物体细节上会有退化。
我的理解是LapLoss更让模型更关注于高频细节的建模，但忽略了低频信息。请问你怎么看待这两种损失的偏好，以及在Practical-RIFE中抛弃了LapLoss的原因？
此外，我试图融合这1）2）两种损失的优点（简单的加权融合），但发现模型貌似难以收敛（加权后的PSNR都低于 单独用其中一种损失），感觉两者之间似乎存在某种对抗？想请问你有何建议。
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
The issue names no file, test, or entry point; begin by locating the training loss definitions and interpolation-training entry point in Practical-RIFE. Compare the Laplacian, L1, and VGG-loss configurations, and consider the issue resolved when the trade-offs and reported convergence behavior receive a documented explanation.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Documentation
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 18/100