README.md

Dataset

Containing files

fullbody_train_charm.csv/fullbody_test_charm.csv

Made from Jan Bre's notebook with DIM=1024, EPOCH=50. We used fullbody_annotations.csv from Jan Bre's dataset as training data.

species.npy/individual_id.npy

Arrays of label encoders used in charmq's pipeline.

pseudo_labels/

List of pseudo labels we created in each round.

Files you need to download from Kaggle

train.csv/sample_submission.csv/train_images/test_images

Competition dataset.

fullbody_train.csv/fullbody_test.csv

Jan Bre's dataset.

train_backfin.csv/test_backfin.csv

Jan Bre's notebook (copy of train.csv/test.csv).

train2.csv/test2.csv

phalanx's dataset. Bboxes of Detic.

yolov5_train.csv/yolov5_test.csv

Awsaf's notebook (copy of train.csv/test.csv).