gabriel-sgama / semantic-superpoint Goto Github PK
View Code? Open in Web Editor NEWOfficial implementation of Semantic SuperPoint (LARS-2022): https://arxiv.org/abs/2211.01098
License: MIT License
Official implementation of Semantic SuperPoint (LARS-2022): https://arxiv.org/abs/2211.01098
License: MIT License
I would like to train the models with other datasets, but have some confusions.
Train_model_frontend_all.py
and Train_model_heatmap_all.py
?*_all_ct.py
in the readme means *_all_and.py
?Could you give an example for a pair of images? Thx.
And what is the difference between superpoint_coco_2017_ML22 and superpoint_coco_ssmall_ML22?
At first I'd like to thank you for this great work !
I want to test your pre-trained models in an implementation that uses .pth files instead of pth.tar. Thus, I tried your script "convert2script.py" in order to convert the weights to a pth model but it raises the following error:
RuntimeError: logs/superpoint_coco_ssmall_ML22/checkpoints/superPointNet_180000_checkpoint.pth.tar is a zip archive (did you mean to use torch.jit.load()?)
As I read, a possible reason is that the version of Pytorch that the model was trained is different with the installed Pytorch version. I'm using Pytorch 1.3.1 since this is the tested version for your implementation.
What version did you use to train the proposed pre-trained models ? Is the Pytorch=1.3.1 the recommended version ?
When I run train4.py, the error shows that no module named "Train_model_heatmap", and I change the front_end_model in "magicpoint_shapes_pair" to "Train_model_heatmap_all", and then the following error occurred:
Traceback (most recent call last).
File "semanticSP/train4.py", line 143, in
args.func(config, output_dir, args)
File "semanticSP/train4.py", line 44, in train_base
return train_joint(config, output_dir, args)
File "semanticSP/train4.py", line 83, in train_joint
train_agent = train_model_frontend(config, save_path=save_path, device=device)
File "semanticSP/Train_model_heatmap_all.py", line 108, in init
self.r = self.config["model"]["real_batch_size"] // self.config["model"]["batch_size"]
KeyError: 'real_batch_size'
What is the problem?
Thanks for the great work!!! I am interested about the implementation of the unc+ct loss for the multi-task learning, and is there any plan to release the code for the unc loss?
There are too few visulaized results in this repo. I have read the paper. Does this model only improve the keypoints' quality? Or this model predict the keypoints' label?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.