data/hockey_pose.py

import json
import os
import datasets
from datasets import Features, Image, Value, Sequence

class HockeyPoseConfig(datasets.BuilderConfig):
    """BuilderConfig for HockeyPose dataset."""
    def __init__(self, **kwargs):
        super(HockeyPoseConfig, self).__init__(**kwargs)

class HockeyPose(datasets.GeneratorBasedBuilder):
    """HockeyPose dataset: Hockey player images with pose keypoints."""

    VERSION = datasets.Version("1.0.0")
    BUILDER_CONFIGS = [
        HockeyPoseConfig(
            name="hockey_pose",
            version=VERSION,
            description="Hockey pose dataset with keypoints",
        ),
    ]

    def _info(self):
        return datasets.DatasetInfo(
            description="HockeyRink keypoint dataset",
            features=Features({
                'image': Image(),
                'image_id': Value('string'),
                'annotations': Sequence({
                    'class_id': Value('int64'),
                    'bbox': Sequence(Value('float32'), length=4),
                    'keypoints': Sequence(
                        Sequence(Value('float32'), length=3),
                        length=56
                    ),
                })
            }),
            supervised_keys=None,
            homepage="https://huggingface.co/datasets/SimulaMet-HOST/HockeyRink",
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "split": "train",
                }
            ),
        ]

    def _generate_examples(self, split):
        """Yields examples."""
        frames_dir = os.path.join(os.path.dirname(__file__), "..", "frames")
        annotations_dir = os.path.join(os.path.dirname(__file__), "..", "annotations")
        
        # Get all image files
        image_files = [f for f in os.listdir(frames_dir) 
                      if f.lower().endswith(('.png', '.jpg', '.jpeg'))]
        
        for idx, image_file in enumerate(sorted(image_files)):
            image_id = os.path.splitext(image_file)[0]
            image_path = os.path.join(frames_dir, image_file)
            annotation_path = os.path.join(annotations_dir, f"{image_id}.txt")
            
            if not os.path.exists(annotation_path):
                continue
                
            # Read YOLO format annotation
            with open(annotation_path, 'r') as f:
                annotations = []
                for line in f:
                    values = [float(x) for x in line.strip().split()]
                    class_id = int(values[0])
                    bbox = values[1:5]  # [x_center, y_center, width, height]
                    keypoints = []
                    
                    # Process keypoints in groups of 3 (x, y, visibility)
                    for i in range(5, len(values), 3):
                        keypoints.append([
                            values[i],      # x
                            values[i + 1],  # y
                            values[i + 2]   # visibility
                        ])
                    
                    annotations.append({
                        'class_id': class_id,
                        'bbox': bbox,
                        'keypoints': keypoints
                    })
            
            yield idx, {
                'image': image_path,
                'image_id': image_id,
                'annotations': annotations
            }