Soccer Ball Tracking Dataset with 1,450 Broadcast Frames

A public sample of 1,450 frames extracted from 10 video clips of professional soccer broadcasts. The dataset is curated for detecting and tracking soccer balls, specifically designed for tiny object detection challenges. It was created by infactory-ai and last updated on 2026-03-02.

Use Cases

• Train tiny object detection models based on the described challenge of tracking small soccer balls.
• Develop tracking algorithms robust to motion blur based on the feature mentioned in the description.
• Benchmark model performance on long-shot scales based on the dataset's stated design.
• Evaluate computer vision models against occlusions based on the dataset's stated purpose.

Strengths

• Contains 1,450 annotated frames extracted from professional broadcast footage.
• Designed specifically for the challenging task of tiny object detection.
• Explicitly targets robustness to motion blur, long-shot scales, and occlusions.

Limitations

• Row count and total dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

infactory-ai

Collection Method

Extracted from professional soccer broadcast video clips.

Freshness

Last updated 2026-03-02 23:01:20; freshness should be verified.