Comments (14)
@carbondriller thankyou for your reply. I am also getting many issues. I was able to build nms i think in colab but there are issues in compatibility of torch and cuda i think. Now I am trying on Jupyter notebook (local runtime). Did you try that approach already ? Also i have pinged you on protonmail for planenet notebook.
from planercnn.
Hello to you all!
I'm currently trying to reproduce this PlaneRCNN on Google Colab too.
But I have exact the same issue ad the OP showed at the beginning
Were you able to reproduce it on Google Colab?
I also tried the versions of CUDA, torch and gcc as @ajithvallabai told, but probaby I did something wrong.
Any advice or tutorials to follow with?
Thanks in advance.
from planercnn.
Similar is mentioned here: #11
Their proposed solution didn't work
from planercnn.
Could you check the compiled files under nms build folder? I'd recommend asking the question on the original PyTorch repo, https://github.com/multimodallearning/pytorch-mask-rcnn.
from planercnn.
I found a potential reason causing this issue. Maybe using PyTorch 0.4.0 could fix the issue, but after compilation, please upgrade back to 0.4.1.
from planercnn.
Hi @carbondriller , were you finally able to do it ?? If possible can you share the google Colab Notebook where you tried using planercnn
from planercnn.
@cgarg-tud I had trouble compiling nms. So I was not able to reproduce planercnn.
If you are wanna look at @art-programmer 's https://github.com/art-programmer/planenet project (predecessor of PlanerRCNN), I can share that Colab Notebook.
I can share my incomplete effort to reproduce PlaneRCNN as well.
Ping me at carbondriller AT protonmail DOT com
from planercnn.
Hello, is your compilation problem solved? @carbondriller thanks!!
from planercnn.
@skq-5233 did u solve this problem
from planercnn.
@carbondriller @cgarg-tud did u solve with colab or else did u install seperately in your machine
from planercnn.
@carbondriller @cgarg-tud , you can use this repos nms it compiling in colab :) https://github.com/jwyang/faster-rcnn.pytorch both pytorch0.4.0 and pytorch 1.0+ version are available
from planercnn.
Guys above is is not a problem. u need to use cuda8+,gcc5+ version, pytorch0.4.0 if you are using this repo(planercnn)
from planercnn.
@ajithvallabai I am using ubantu 18.04,gcc 7.5, cuda 10.1,pytorch 1.7 I am still getting the warning when i do python build.py same as mentioned above
I have tried with pytorch 0.4.1,0.4 did help..
If i go ahead and run evaluate.py I am getting ImportError: /home/sujunyu/planercnn-master/nms/_ext/nms/_nms.so: undefined symbol: __cudaRegisterFatBinaryEnd
Since I have nvidia 2080 ti I have set the arch = sm75
If there is any steps to solve this?
from planercnn.
@Karthikaddagadderamesh go with cuda8.0, torch 0.4.0, gcc 5+ , build nms,roi align , if you face issues after this in evaluate, update to 0.4.1 and then try it will work
from planercnn.
Related Issues (20)
- how can i get the normal of planes for each frame extracted from /scene_id/scene_id.sens? HOT 2
- Pretrained model links are not working HOT 3
- Nothing happened with nvcc HOT 6
- FileNotFoundError: [Errno 2] No such file or directory: '../../Data/ScanNet//scannetv2-labels.combined.tsv' HOT 5
- Pretrained model links are not working now HOT 2
- Torch summary on the model
- The ground truth provided by PlaneRCNN has large errors! HOT 2
- After training scannet from scratch while inferencing the trained model facing error HOT 1
- Plane parameter
- Unable to compile NMS and ROI-Align on RTX 3080 HOT 3
- Question about Training Data
- Docker image HOT 2
- Keep the order of original image files while detecting planes
- plane_XYZ coordinate system
- cuda 8.0 on Ubuntu 18.0.4
- Plane segmentation evaluation experiment on ScanNet
- Trained Models
- Problem on download of scannet plane annotation
- How to find the plane normal? HOT 1
- How to transfer module from 'pth' to βonnxβοΌ eg: checkpoint_refine.pth to checkpoint_refine.onnx
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
π 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. πππ
-
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 β€οΈ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from planercnn.