Comments (1)
If you want to change to 400*400.
datasets.py: (1) in get_imgs() change imsize to [0,400]. (2) change my_crop_width into a custom value, which I will explain how to calculate later. If this is not working, then I recommend you first to check what is output from our current dataloader and write your own dataloader.
train_first_stage.py: (1) change values of self.patch_stride, self.n_out, self.recp_field, such that
recp_field+(n_out-1)*stride = the custom value. These values are tied with the background discriminator. For example, we are using a patch discriminator with recp=34, n_out:24, stide:4, thus our custom value is 126. You need to design your a patch-level background discriminator, and get their recp_field, n_out and stride. (2) change number 126 in L219 and L221 into your custom value.
model_train.py : Change architecture of all models: genrerator, background/parent/child discriminator, encoder, bi_discriminator, feature_extracotr such that they can process or generate 400px images. Note that background discriminator architecture determines the values mentioned above.
train_second_stage.py: Since you will modify parent generator, thus you need to change argument in L45, and L49. The current number 16 means number of channels of parent feature(the one fed into child generator), once you change architrcure replace 16 with your number of channels.
from mixnmatch.
Related Issues (20)
- question about eval result and re-train result HOT 2
- why not initialization the weight of BD_Networks? HOT 1
- Requirements HOT 2
- Transfer Learning on MixnMatch
- How to print EG_Loss? HOT 1
- Result is very bad on custom dataset HOT 6
- image resize HOT 6
- Parameter details HOT 1
- Colab version please? HOT 2
- Bounding Box HOT 3
- where the trained modles? HOT 2
- Unable to download pretrained models HOT 1
- Unable to setting up the dataset HOT 2
- What is the Bounding box? HOT 3
- Large amount of data is needed? HOT 2
- About pre-training model
- mutual information discriminator
- How can I train and create a new model for custom dataset?
- Related to the bounding box annotation.
- why multiple 4 ?? HOT 2
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 mixnmatch.