1adrianb / 1adrianb/face-alignment
Parallelizing on the CPU
- Dominant language
- Python
- Stars
- 7.5k
- Forks
- 1.4k
- PR merge metrics
- No merged PRs in 30d
Description
Is it possible or has anyone tried to parallelize face-alignment on the CPU? I tried to do so with `from multiprocessing import Pool`, but I ran into [this pytorch issue](https://github.com/pytorch/pytorch/issues/18106).
I'm assuming the thing to do is to just run on a GPU, but I thought I would put it out there in case someone already knows a way to make multiprocessing work.
Contributor guide
No contributing guide indexed for this repository
Research direction
Look at the face-alignment library's inference code to understand the model loading and prediction flow. Investigate the PyTorch multiprocessing issue linked, and explore alternatives like threading, joblib, or using PyTorch's own DataLoader with num_workers. Check if there are existing examples of CPU parallelization in similar PyTorch-based vision libraries.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- performance
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100