About rotation of the data augmentation
Closed this issue · 2 comments
taokong commented
Hi, @bearpaw , very clear code!
In /pose/datasets/mpii.py line 95
r = torch.randn(1).mul_(rf).clamp(-2*rf, 2*rf)[0] if random.random() <= 0.6 else 0
The rotation angles are between [-60, 60]. It is not consistent with the Hourglass work and your paper(Multi-Context Attention for Human Pose Estimation) [-30, 30]. I also see similar codes at pose-attention .
Does larger angle variation help training of pose detection?
bearpaw commented
Sorry that I don't quite remember the reason for this setting. Maybe it's a mistake or maybe it is following a history version of pose-hg-train.
I believe the performances of different settings should be very similar. You can have a try.
taokong commented
Thanks