Comments (5)
Yes, the inputs to both models are transformed with different augmentations. To calculate the loss between the student and teacher model, we have to map the coordinates to the same space. We use the transform_matrix
to record the augmentation information. And then we are able to align inputs auged with different augmentations with the matrix.
from softteacher.
Yes, the inputs to both models are transformed with different augmentations. To calculate the loss between the student and teacher model, we have to map the coordinates to the same space. We use the
transform_matrix
to record the augmentation information. And then we are able to align inputs auged with different augmentations with the matrix.
thanks! clearly understood
from softteacher.
I have another question- can't seem to find it in the paper. At inference time, is the teacher model or student model used for inference? Thanks.
from softteacher.
@cnmckee In our paper and provided code, we use the student model for inference. However, in some cases, the teacher performs somewhat better than the student. So just for advice, you can try them both to see which is better.
from softteacher.
@cnmckee In our paper and provided code, we use the student model for inference. However, in some cases, the teacher performs somewhat better than the student. So just for advice, you can try them both to see which is better.
Makes sense- thanks for the speedy reply!
from softteacher.
Related Issues (20)
- lr_config set up according to max iter?
- Config files for evaluating the provided models
- 关于PseudoSamples
- colab
- Cascade RCNN with Soft Teacher does not run: _bbox_forward() missing 1 required positional argument: 'rois'
- 在compute_pseudo_label_loss()的时候已经对teacher生成的伪标签做了坐标变换,为什么在unsup_rcnn_cls_loss()的时候还要变换一次呢?
- 在代码中看到TrimapMasks这个类,是做半监督实例分割要用的吗
- How to calculate this parameter "epoch_length=7330" HOT 1
- sup_loss_bbox: nan and loss:nan HOT 1
- for instance segmentation
- Unable to train using 1 GPU HOT 5
- Welcome update to OpenMMLab 2.0
- ModuleNotFoundError: No module named 'ssod' HOT 1
- MMDetection v3.0
- sup:unsup
- 如何处理的student多次forward 的loss HOT 2
- by mistake
- how I can use this in dota datasets?
- if the batch is bigger , the training time is longer? HOT 1
- Environmental configuration
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 softteacher.