Comments (2)
Hello,
The offsets variable controls the way we generate sliding windows (whether we start from minimum coordinates or minimum coordinates plus a half block size offset). We actually inherit this setting from PointCNN implementation. The reason why we use these offsets is that the border regions of sliding windows might not be processed properly if we have just zero offset (for example, a chair might be separated into two parts if it lies on the border).
On the test data, it is OK to generate sliding windows with one offset or two. The later is adopted by PointCNN and us, and achieves better performance.
Best,
Haotian
from pvcnn.
I'm closing this issue due to inactivity. Please feel free to reopen it if the problem has not been resolved.
from pvcnn.
Related Issues (20)
- RuntimeError: Error building extension '_pvcnn_backend': HOT 3
- CUDA kernel failed : an illegal memory access was encountered HOT 2
- How to convert custom op into tensorrt HOT 1
- Why use the code x = x - (minx + args.block_size / 2) in data/s3dis/prepare_data.py? HOT 2
- Two questions about feature extraction HOT 2
- error when running "from modules.functional.backend import _backend" HOT 11
- Function TrilinearDevoxelizationBackward returned an invalid gradient at index 0 - got [4, 192, 32, 32, 32] but expected shape compatible with [4, 192, 34, 34, 34]
- voxelization part HOT 1
- Some issues about feature propessing. HOT 4
- Data Pre processing Issue. HOT 4
- Code about S3DIS visualization HOT 1
- About ModuleNotFoundError: No module named 'config.shapenet' HOT 1
- Can I use just one step of voxelization and devoxelization operator to voxelize and devoxelize the point cloud? HOT 2
- Pretrained Models Page Not Found HOT 3
- Voxel features are all zero HOT 3
- ImportError: DLL load failed while importing _pvcnn_backend: 找不到指定的模块。 HOT 2
- Model Inference HOT 1
- vox_coords = torch.round(norm_coords).to(torch.int32) HOT 1
- nvcc error : 'cudafe++' died with status 0xC00000FD (STACK_OVERFLOW) HOT 1
- Question about voxelization modul (compares to spvoxelize in torchsparse) HOT 1
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 pvcnn.