PRBonn
关于使用rangenet的问题 (Problem when using rangenet)
您好，我在使用您的代码时，infer这步出现了如下问题： Couldn't load backbone, using random weights. Error: ../../../data/model_rangenet_residual_1/backbone is a zip archive (did you mean to use torch.jit.load()?) Couldn't load decoder, using random weights. Error: ../../../data/model_rangenet_residual_1/segmentation_decoder is a zip archive (did you mean to use torch.jit.load()?) Couldn't load head, using random weights. Error: ../../../data/model_rangenet_residual_1/segmentation_head is a zip archive (did you mean to use torch.jit.load()?)
Infering in device: cuda OMP: Warning #191: Forking a process while a parallel region is active is potentially unsafe. OMP: Warning #191: Forking a process while a parallel region is active is potentially unsafe. OMP: Warning #191: Forking a process while a parallel region is active is potentially unsafe.
但之后还可以生成prediction文件，并且在评估iou时，08序列res=1时为0.005（这个估计是错误的）您知道怎么回事吗
刚刚看了一下我的torch：
import torch
print(torch.version) 1.1.0 import torchvision print(torchvision.version) 0.2.2
然后咱们的requirement是： torch==1.1.0 tensorflow==1.13.1 vispy==0.5.3 torchvision==0.2.2.post3
应该是一致的把
您好，我使用这篇文章https://zhuanlan.zhihu.com/p/454415689，更新了一下pytorch版本，现在已经成功运行，就是运行的时候使用了cpu，您知道怎么切换回cuda吗 model folder exists! Using model from ../../../data/model_rangenet_residual_1 Sequences folder exists! Using sequences from ../../../data/sequences parsing seq 08 Using 4661 scans from sequences [8] Using DarknetNet53 Backbone Depth of backbone input = 6 Original OS: 32 New OS: 32 Strides: [2, 2, 2, 2, 2] Decoder original OS: 32 Decoder new OS: 32 Decoder strides: [2, 2, 2, 2, 2] Total number of parameters: 50372739 Total number of parameters requires_grad: 50372739 Param encoder 40585792 Param decoder 9786080 Param head 867 Successfully loaded model backbone weights Successfully loaded model decoder weights Successfully loaded model head weights
Cleaning point-clouds with kNN post-processing kNN parameters: knn: 5 search: 5 sigma: 1.0 cutoff: 1.0 nclasses: 3
Infering in device: cpu Infered seq 08 scan 000000.label in 10.995035409927368 sec Infered seq 08 scan 000001.label in 9.829725742340088 sec