Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Added the dataset wrapper for the Human3.6m. Original implementations have it return a$32 \times 1$ pose for the 2d case and a $48 \times 1$ pose for the 3d case. Therefore, every implementation returned only one pose, making the entire dataset having around 2 million poses, which makes sense.
However, in our case we want a sequence of poses so we omit the last part where the sequence is "flattened", meaning every time step of the sequence counts as an individual sample. We only maintain the 3d skeleton (we only care about 3d skeleton forecasting) and we end up with 600 variable sequence length skeleton motions.
See
src/skelcast/data/dataset.py
for the addition.Seems like that it can work with the
NTURGBDCollateFnWithRandomSampledContextWindow
right away without modifications... not tested yet.