Comments (9)
@JANEKKO Hi,
First, create a dataset class for your own data in the form of ShapeNet-55 dataset class here, if you align the output for __getitem__
function with other dataset in this repo, you can save the time to modify the code in runner.py
.
The config file for dataset class should be define in the cfgs/dataset_configs
. For example, ShapeNet55_meta_config. It will be loaded and used when initialize the dataset class if you specify the _base_
in model config, like PoinTr_on_ShapeNet55.
Second, you can create a model config folder. For example. a config folder for ShapeNet55 here. And put the meta config for PoinTr into this folder. (Do not forget to modify the dataset
config in this yaml file)
Last, you can test the pretrain PoinTr by change the config file to the newly create one.
from pointr.
@blukaz Have you initialized your dataset class in the `datatsets/__init__.py'? For example, here
from pointr.
Here is are some additional steps for others who are trying to train the PoinTr model on their own (custom) dataset:
- add this line in the ".datasets/init.py" script:
import datasets.CUSTOMDATASETNAME
- add the support for your dataset in the ".tools/runner.py" script: take a look at line 82 (def run_net) and line 178 (def validate) in the code
- optional: if you need support for ".ply" files then you have to add this in ".datasets/io.py" script
- important: don't forget to center and normalize your point cloud data
from pointr.
from pointr.
@JANEKKO feel free to reopen the issue if you find something difficult to cope with.
from pointr.
Hi yuxumin,
Thanks for the insights, but I get the following error:
Traceback (most recent call last): File "main.py", line 68, in <module> main() File "main.py", line 64, in main run_net(args, config, train_writer, val_writer) File "/content/drive/MyDrive/ColabNotebooks/PoinTr_setup/PoinTr/tools/runner.py", line 17, in run_net (train_sampler, train_dataloader), (_, test_dataloader) = builder.dataset_builder(args, config.dataset.train), \ File "/content/drive/MyDrive/ColabNotebooks/PoinTr_setup/PoinTr/tools/builder.py", line 14, in dataset_builder dataset = build_dataset_from_cfg(config._base_, config.others) File "/content/drive/MyDrive/ColabNotebooks/PoinTr_setup/PoinTr/datasets/build.py", line 16, in build_dataset_from_cfg return DATASETS.build(cfg, default_args = default_args) File "/content/drive/MyDrive/ColabNotebooks/PoinTr_setup/PoinTr/utils/registry.py", line 148, in build return self.build_func(*args, **kwargs, registry=self) File "/content/drive/MyDrive/ColabNotebooks/PoinTr_setup/PoinTr/utils/registry.py", line 282, in build_from_cfg f'{obj_type} is not in the {registry.name} registry') KeyError: 'SIW is not in the dataset registry'
It seems that my custom dataset doesn't get loaded into the registry:
Registry(name=dataset, items={'KITTI': <class 'datasets.KITTIDataset.KITTI'>, 'PCN': <class 'datasets.PCNDataset.PCN'>, 'ShapeNet': <class 'datasets.ShapeNet55Dataset.ShapeNet'>})
I tried to debug it in the registry.py script, but I couldn't figure out what the problem is. In line 270 it merges the new config successfully and then gets kicked out at 278.
Note: I've added the dataset config file, model config file and the python script in the datasets folder.
Thanks in advance!
from pointr.
@yuxumin Thanks! That did the job
from pointr.
Here is are some additional steps for others who are trying to train the PoinTr model on their own (custom) dataset:
- add this line in the ".datasets/init.py" script:
import datasets.CUSTOMDATASETNAME
- add the support for your dataset in the ".tools/runner.py" script: take a look at line 82 (def run_net) and line 178 (def validate) in the code
- optional: if you need support for ".ply" files then you have to add this in ".datasets/io.py" script
- important: don't forget to center and normalize your point cloud data
Excuse me, have you tried training this model on a dataset of large outdoor buildings
from pointr.
from pointr.
Related Issues (20)
- Pretrained AdaPoinTr models HOT 3
- The question about the code of GRNet in this repository. HOT 2
- No module named 'PoinTr' found HOT 1
- 大佬你好,注意到你的shapenet2pc只有采样到8192分辨率,那残缺点云是如何制作的呢
- Training error: ModuleNotFoundError: No module named 'pointnet2_ops' HOT 2
- Question about the query ranking.
- Training PoinTr on Custom Dataset HOT 4
- How to get the FLOPs and Throughput like table.9 in the paper?
- 测试完PCN数据集只输出了一个log HOT 3
- Some questions about KITTI
- In KITTI test, the generated point cloud is not base on input HOT 3
- Works on CUDA11.1 but not on CUDA11.8 HOT 1
- 训练权重
- SSC code
- 2 errors detected in the compilation of "gridding.cu"
- How the foreground region is determined for downstream object detection tasks
- 有关训练出来的数据集的问题 HOT 1
- 关于自制kitti数据集训练的问题
- 使用预训练模型PCN
- Version of ShapeNet
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 pointr.