search

oarriaga / paz

Hierarchical perception library in Python for pose estimation, object detection, instance segmentation, keypoint estimation, face recognition, etc.
https://oarriaga.github.io/paz/
MIT License
617 stars 102 forks source link

Fer2013.csv dependency not found #309

Closed benjaminclay01 closed 1 year ago

benjaminclay01 commented 1 year ago

When trying to run the examples/face_classification/pipelines.py or train.py, program cannot find fer2013.csv. After investigating, I have found that paz/datasets/ferplus relies on having a path to a dataset that is not contained within the repository. As such, I have started to look online for the ferplus2013 database, I have found the database online, but am unaware of how to allow PAZ to see the csv in question.

benjaminclay01 commented 1 year ago

I am aware this is more requesting aid than reporting an issue; I apologize for the inconvenience.

oarriaga commented 1 year ago

Hi @benjaminclay01 no problem! Feel free to ask us here. Your question can also be useful for someone in the future with a similar situation. Unfortunately we can't provide the dataset since it violates it's LICENSE.

  1. You can read the documentation of the data manager class [FerPlus].(https://github.com/oarriaga/paz/blob/b3e3758907192c69aa5fdc7f27ea7ac031c8b2fb/paz/datasets/ferplus.py#L12).
  2. This data manager hints (we should do a better way at communicating this!) that one needs to download the dataset from Kaggle.
  3. After the dataset is downloaded you should only need to provide the path to the csv file, as indicated in the documentation.

P.S. If you want you are free to send us a README.md inside paz/examples/face_classification e.g. paz/examples/face_classification/README.md that describes this process in a high level. You can also check this example using a kaggle dataset for programmatic solution.

I hope this helps!