Comments (6)
The main code you would need to modify is under the ./lib/datasets folder. In that folder, you can copy one of the {KITTI, bdd100k, cityscapes} code and change according to your needs. Afterward, you would need to add that dataset into ./lib/datasets/factory.py and give it a name. Then you should be able to specify your dataset in one of the scripts in ./experiments/scripts. These are the major changes to get your dataset up and running.
from da_detection.
The main code you would need to modify is under the ./lib/datasets folder. In that folder, you can copy one of the {KITTI, bdd100k, cityscapes} code and change according to your needs. Afterward, you would need to add that dataset into ./lib/datasets/factory.py and give it a name. Then you should be able to specify your dataset in one of the scripts in ./experiments/scripts. These are the major changes to get your dataset up and running.
Which pre-trained weights will be used to customize the data set in the first phase of the training network? If you can reply, thank you very much.
from da_detection.
Which pre-trained weights will be used to customize the data set in the first phase of the training network? If you can reply, thank you very much.
Hi @life97, have you figured this out?
from da_detection.
According to the author's process, the model can be run through. I used this model to train the network on a custom domain-adapted scene dataset. The network training is divided into two stages, the first stage is the source domain to the intermediate domain, and the second stage is the intermediate domain to the target domain. The CycleGAN is jointly trained by the source domain and the target domain to synthesize the intermediate domain images, and the weights of the intermediate domain images are the discriminator outputs of the trained CycleGAN. It's been a long time, I probably remember it like this, I hope it helps you.
from da_detection.
@life97 thank you for your reply. I have generated my custom dataset from PyTorch CycleGAN, but still couldnt understand which pre-trained weights to use with it. Would it be the weights from CycleGAN?
from da_detection.
@life97 thank you for your reply. I have generated my custom dataset from PyTorch CycleGAN, but still couldnt understand which pre-trained weights to use with it. Would it be the weights from CycleGAN?
Step 1: Modify the return value of the discriminator in H:\19xyy\project\DA_detection-master\PyTorch-CycleGAN\models.py: (change from line 92 to line 93)
Step 2: Modify test.py as needed
Mainly modify PyTorch-CycleGAN: datasets.py and test.py
I changed it like this, the general idea is this, I'm not sure if this is right, you can try~
If there are any mistakes, please correct me, I hope it helps you~
from da_detection.
Related Issues (20)
- could you explain how to train full network step by step? HOT 3
- Reported Accuracy for cross camera adaptation on which dataset HOT 3
- What are motor and bike classes for Cityscapes HOT 1
- About Cross Camera Adaptation HOT 1
- image size when generating from pre-trained weight HOT 3
- No module named 'later_utils.roi_align._ext.crop_and_resize._crop_and_resize' HOT 4
- About One stage model HOT 1
- About Train CycleGAN HOT 3
- the labels of the synthetic images,how to get it ? HOT 1
- 文献公式2咨询
- 文献公式2咨询
- AttributeError: 'EasyDict' object has no attribute 'FPN' HOT 2
- Evaluation AP
- How to generate score json file HOT 1
- loss nan
- Can I deploy the project under the windows system? HOT 1
- Error
- Foggy cityscapes dataset
- module 'layer_utils.roi_align._ext.crop_and_resize' has no attribute 'crop_and_resize_gpu_forward'
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 da_detection.