Sign in to view source links and access this dataset
Description
Keremberke's NFL Object Detection dataset contains 9,947 images split into training, validation, and test sets. The images are annotated for object detection with labels related to football helmets, including 'helmet', 'helmet-blurred', 'helmet-difficult', 'helmet-partial', and 'helmet-sideline'. The dataset was last updated on the Hugging Face platform on January 29, 2023.
Use Cases
Train object detection models to locate football helmets in game footage based on the described annotation labels.
Benchmark model performance on challenging visual conditions like blurred or partially occluded helmets.
Develop sideline-specific detection systems for specialized camera angles mentioned in the label set.
Fine-tune pre-trained models for sports safety applications by detecting helmet presence and condition.
Strengths
Dataset provides 6,963 images for training, 1,989 for validation, and 995 for testing, offering a substantial split for model development.
Annotations include five distinct, scenario-specific labels for football helmets, which likely aids in training models for varied real-world conditions.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Last updated 2023-01-29 12:37:17; freshness should be verified for current applications.
Provenance
Source
keremberke via Hugging Face Datasets.
Freshness
2023-01-29
License is unknown; users should verify terms before use. Loading requires the `datasets` Python library.