Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Flip_item is not applied during training

See original GitHub issue

Dear authors, thanks for the great work!

I noticed that parameter flip_item passed into the model is set to None during train() but is popped from is_flipped key during validate_gt()

In train: https://github.com/Jeff-sjtu/HybrIK/blob/b8cfeeb7df2d7c8024865751cc17664a8272557b/scripts/train_smpl.py#L57

in validate_gt: https://github.com/Jeff-sjtu/HybrIK/blob/b8cfeeb7df2d7c8024865751cc17664a8272557b/scripts/train_smpl.py#L128-L130

When flip_item is set toNone, the phi, leaf, shapes and uvd_coordinates will not be flipped during training even though the image and keypoints3d might be flipped during augmentations. https://github.com/Jeff-sjtu/HybrIK/blob/e10afb149e10cf721b619b12eb81f02b7ad3c4e0/hybrik/models/simple3dposeBaseSMPL24.py#L298-L325

I would also like to confirm if flip_item is indeed set to None for all training? And if yes, why is flipping of phi, shapes and uvd_coord not applied during training, but applied during validation?

I am rather confused regarding this procedure, and hope to get some clarification on this. Thank you!

Issue Analytics

State:
Created 2 years ago
Comments:7 (3 by maintainers)

Top GitHub Comments

2reactions

Jeff-sjtucommented, Nov 2, 2021

Hi @pangyyyyy, I think the bug is fixed now!

0reactions

pangyyyyycommented, Nov 3, 2021

That’s great! Thank you so much!

Top Results From Across the Web

Vision augmentation - fastai

Transforms to apply data augmentation in Computer Vision. ... is looked for to decide if the transform is executed or not). ... FlipItem...

fastai2/augment.py at master - GitHub

Temporary home for fastai v2 while it's being developed - fastai2/augment.py at master ... if resize_to is not None: x = x.resize(resize_to, resize_mode)....

Inference With fastai - Model Saving, Loading, and Prediction

In this tutorial I cover how to use fast.ai for inference, ... We don't need FlipItem , as it's not applied on the...

fastai.pdf

It is the default weight decay used when training the model. wd_bn_bias. It controls if weight decay is applied to BatchNorm layers and...

DINO ImageWang Tutorial | self_supervised - GitHub Pages

Note: This notebook demonstrates how to use DINO callback with a single GPU. First import fastai for training and other helpers, you can...