Luke Chang comments

Results 80 comments of


                                            Luke Chang

make lassopcr predict option more clear in documentation

added to docstring

Missing frames in detect_video results

Hi @Ben-FCC , apologies for the issues you are having. We are currently in the process of updating both how batch processing is being done and error handling, which I...

Face detector performance is low on small images.

Good point @mperreir. @ejolly as we continue to finish up https://github.com/cosanlab/py-feat/pull/133, we may want to now allow specific modules to be turned off in the main image/video detector methods. I...

How to use au_model="rf" and emotion_model = "rf" in version 0.4

hi @ritika24-s , We discovered an issue with the output of the RF model, which we haven't had time to fully debug or retrain yet. We disabled it in version...

Different AU values when passing to detector.detect_image a single image or a batch of images

Hi @VaianiLorenzo, thanks for sharing this. It is definitely not expected. We are working on a major refactor to the code base, which I suspect will address this issue (https://github.com/cosanlab/py-feat/pull/133)....

Feature Request: Add tqdm for verbose processing

This is now implemented in the soon to be released new version that uses pytorch dataloaders (https://github.com/cosanlab/py-feat/pull/133).

Batch_size issue in latest update

We are completely refactoring how batching is done, which should address this issue. It's done, but not working with all detectors yet https://github.com/cosanlab/py-feat/pull/133.

Img2pose method can only get yaw, roll and pitch at last. Can 6D parameter be included in the result as in the original paper or Openface

Hi @110wuqu, thanks for the suggestion. We will look into adding these extra parameters. We initially chose not to include them to avoid confusion with the bounding box information. @ejolly...

add support for m1 GPU and refactor code to speed up detection with GPUs

Doesn't seem like this is available yet on github.

add support for m1 GPU and refactor code to speed up detection with GPUs

Updated Face detection module 1) Now works with M1 GPUs. 2) Can now accommodate images from opencv, torchvision, or PIL 3) Cleaned up code