search

ShiqiYu / OpenGait

A flexible and extensible framework for gait recognition. You can focus on designing your own models and comparing with state-of-the-arts easily with the help of OpenGait.
720 stars 165 forks source link

IndexError: list index out of range #243

Open DrShanUllah opened 5 days ago

DrShanUllah commented 5 days ago

python 3.9.0 Pytorch : 2.1.1+cu118

I was able to run Gaitbase basic code on gait3d using silhouettes. ####################Silhouettes based models are working for me ############## I am able to run [successful] model\gait\output\Gait3D\Baseline\GaitBase_DA\checkpoints\GaitBase_DA-60000.pt This is successful.

############### Skeleton-based models are not working for me ############### However, I am unable to run any skeleton-based model.

However, I tried to run a skeleton-based algorithm on the same environments. But I was unsuccessful.

################################ Error 1 ############################## 1) model\gait\output\GREW\SkeletonGaitPP\SkeletonGaitPP\checkpoints\SkeletonGaitPP-180000.pt Error: skeletongait++.py", line 74, in inputs_pretreament for pose, sil in zip(pose_sils[0], pose_sils[1]): IndexError: list index out of range

################################ Error 2############################## 2) \model\gait\output\Gait3D\DeepGaitV2\SkeletonGait\checkpoints\SkeletonGait-60000.pt Error: torch\nn\modules\conv.py", line 456, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: Given groups=1, weight of size [64, 2, 3, 3], expected input[138, 1, 64, 44] to have 2 channels, but got 1 channels instead

For vidoe processing, do i need to follow any other step?

DrShanUllah commented 3 days ago

It seems I missed processing the skeleton pose, which is why it throws an error requiring two channels. Regarding silhouettes, as paddle-seg is recommended, do we have a recommended repository for HRNet ? Do you have complete code processing for skeleton + silhouette in case we have input video? as current repository assumed that we have already processed silhouettes and already extracted skeletons?