XiandaGuo/OpenStereo

Issues

• 使用自制数据集，训练时会出现loss为nan的情况。

  #150 opened 2 months ago by t973288913
  1

• Plans to upload StereoBase weights to google drive?

  #154 opened 2 months ago by MalekWahidi
  1

• Plan about model deployment

  #92 opened 6 months ago by Daa98
  2

• 关于复现StereoBase

  #94 opened 8 months ago by shengwenbo125
  2

• KITTI benchmark sota is based on which model?

  #79 opened 10 months ago by ziming-liu
  2

• Are the code and weights for StereoBase available?

  #97 opened 7 months ago by MalekWahidi
  2

• where is the Stereobase trainning model code and yaml configure file in this project?

  #149 opened 2 months ago by RDXiaoLu
  2

• about kitti 2012 submission. you provite test_kitti.yaml product images 1242 375 did't submission. should be 1241 376

  #152 opened 2 months ago by RDXiaoLu
  0

• How to use ckpt continue train?

  #151 opened 2 months ago by RDXiaoLu
  1

• 使用IRS数据训练，前面训练正常，但是到第9epoch时output为nan，不知道怎么排查

  #141 opened 3 months ago by yangning6103
  2

• about EPE

  #140 opened 3 months ago by Redreadblack
  1

• 作者你好，我使用openstereo复现lightstereo在第14轮出现了梯度消失，这个怎么办！

  #130 opened 3 months ago by dididichufale
  4

• 使用test_kitti.py加载LightStereo-L-KITTI.ckpt权重测试kitti2012和kittie2015的效果不理想，这是可能是什么原因呢

  #148 opened 2 months ago by luoxieyang
  0

• Infering on different data

  #139 opened 2 months ago by danielvais
  1

• 你好，我想推理视频流

  #138 opened 2 months ago by Disembark1
  1

• question about having large weights after training

  #147 opened 2 months ago by 13185742215
  0

• 如何使用两个数据集训练

  #142 opened 2 months ago by 13185742215
  1

• 其他数据集的训练配置文件

  #146 opened 2 months ago by 13185742215
  1

• Why LightStereo has poor result for Sky area

  #136 opened 3 months ago by wattanakorn495
  1

• 关于lightStereo里面采样mobilenetv2的训练过程

  #134 opened 3 months ago by Hlll-huang
  3

• question about the code of context network and upsampling

  #145 opened 2 months ago by zhaoxin111
  0

• Understanding the outputs

  #132 opened 2 months ago by rachitagarwal2603
  1

• igev 相较于原官方的实现推理一对图像，耗时是官方的10倍左右？

  #126 opened 2 months ago by nmber5
  2

• 无法推理

  #118 opened 2 months ago by Hlll-huang
  5

• Questions about the combination of augmentations.

  #143 opened 2 months ago by Windsrain
  0

• where is the Stereobase trainning model code and yaml configure file in this project?

  #137 opened 3 months ago by Brian-YANGGY
  1

• 你好，关于步态识别grew Competition的问题

  #135 opened 3 months ago by ZunxiaoXu
  1

• 关于lightStereo里面采样mobilenet

  #133 opened 3 months ago by Hlll-huang
  0

• Hello author, thank you for your work！ I have a problem.

  #125 opened 4 months ago by asd9697
  2

• Training model using a model ckpt

  #129 opened 4 months ago by danielvais
  1

• Infer error

  #124 opened 4 months ago by paleriderrrr
  9

• 前辈您好，想请教一个问题，谢谢

  #123 opened 5 months ago by flyxuexi
  0

• Quick question on inference speed of LightStereo

  #120 opened 5 months ago by Jm7752
  2

• tools/eval.py has no pretrained_model access

  #105 opened 5 months ago by MichaelWangGo
  1

• ImportError: cannot import name 'deform_conv_cuda' from partially initialized module 'stereo.libs.AANet.deform_conv'

  #111 opened 5 months ago by lvruizhen
  1

• Why the performances of cross-domain evaluation are not same as original paper?

  #115 opened 5 months ago by lqxisok
  1

• Why is the learning rate scheduler in the config file of the model zoo MultiStepLR?

  #113 opened 6 months ago by yellowYuga
  1

• Hi what does "OOM" means in your paper, please?

  #106 opened 6 months ago by Elenairene
  1

• The size of tensor a (156) must match the size of tensor b (155) at non-singleton dimension 4

  #100 opened 6 months ago by lixxxxj
  1

• 使用kitti数据集训练psmnet时似乎有问题，使用train.py训练时打印出的eval数据集的epe过低（一点几，和loss相近），使用eval.py打印出的eval数据集的epe过高（三十几），这与许多论文的结果不一致

  #101 opened 6 months ago by XianghuiFan
  1

• Middlebury dataset download issue

  #96 opened 7 months ago by zhangzscn
  1

• Inference on single image

  #88 opened 7 months ago by saqib736
  1

• The desgin is not inference friendly.

  #98 opened 7 months ago by luxunxiansheng
  1

• Mistake in paper table7

  #95 opened 8 months ago by Guptajakala
  1

• Is this article a review and finishing work of stereo matching？

  #87 opened 8 months ago by progressforever
  1

• The training strategies

  #83 opened 9 months ago by zhuxing0
  1

• Which pretrained model exactly is stereo base?

  #81 opened 9 months ago by Rong-Tao
  1

• Configuration file for the DrivingStereo dataset.

  #84 opened 9 months ago by Tobyzai
  1

• There are some issues with the configuration and reproduction of Raft-stereo

  #76 opened a year ago by qylen
  1

• sttr, epe is tool large

  #62 opened a year ago by arminghz
  2