infactory-ai/camera-detection

Soccer Camera Detection Dataset

A curated dataset for classifying camera characteristics in broadcast soccer video, annotated with 9-class shot scale (CineScale-aligned) and 5-class camera angle (CineScale2-aligned).

Shot Scale Examples

Left: MCU Medium Close-Up — head & shoulders  |  Center: LS Long Shot — full body  |  Right: ELS Extreme Long Shot — wide pitch view

Camera Angle Examples

Left: HIGH High Angle — elevated broadcast camera  |  Center: NEU Neutral — pitch-level eye height  |  Right: LOW Low Angle — camera below subject

Dataset Description

This public sample consists of 650 frames extracted from 10 video clips of professional soccer broadcasts spanning 2013–2025. Each frame is annotated across two orthogonal dimensions:

• Shot Scale — how much of the subject is visible (9 classes from Extreme Close-Up to Extreme Long Shot)
• Camera Angle — the vertical orientation of the camera relative to the subject (5 classes from Overhead to Low Angle)

This is a representative sample of a larger dataset available for licensing. The full dataset covers hundreds of matches across multiple seasons and broadcast conditions — suitable for model training, fine-tuning, and research. Contact hello@infactory.ai for licensing details.

Shot Scale Distribution

Class	Frames	%
MLS	167	25.7%
ELS	159	24.5%
LS	118	18.2%
MS	85	13.1%
MCU	50	7.7%
FS	43	6.6%
IS	14	2.2%
CU	8	1.2%
ECU	6	0.9%

Camera Angle Distribution

Class	Frames	%
NEU	369	56.8%
HIGH	252	38.8%
LOW	29	4.5%

Source Data

• Domain: Professional Soccer Broadcasts (Serie A via Infront Italy)
• Resolution: 1280×720 (720p)
• Annotation Style: Per-frame JSON
• Labeling Method: Proprietary model + human verification (100% verified)

Dataset Structure

infactory-ai/camera-detection/
├── README.md
├── metadata.csv
├── dataset_info.json
├── taxonomy/
│   ├── shot_scale.json
│   ├── camera_angle.json
│   └── mappings.json
├── schema/
│   └── annotation_schema.json
└── data/
    └── frames/
        └── {uuid}_{frame_number}.jpg

Metadata Fields (metadata.csv)

Field	Type	Description
frame_id	string	Unique frame identifier (UUID)
source_video_id	string	UUID of the source video clip
frame_number	int	Frame index in the source clip
timestamp_sec	float	Timestamp in seconds
shot_scale	string	Shot scale label (ECU, CU, MCU, MS, MLS, LS, ELS, FS, IS)
shot_scale_code	int	Numeric code (0–8)
camera_angle	string	Camera angle label (OVH, HIGH, NEU, LOW, DUTCH)
camera_angle_additional	string	Additional angle labels if multi-label (pipe-separated)
subject_type	string	Primary subject type
camera_position	string	Physical camera position
confidence	float	Model confidence before human verification
human_verified	bool	All frames are human-verified

Usage

Loading with Hugging Face Datasets

from datasets import load_dataset

dataset = load_dataset("infactory-ai/camera-detection")

Loading with pandas

import pandas as pd

df = pd.read_csv("metadata.csv")
print(df["shot_scale"].value_counts())
print(df["camera_angle"].value_counts())

Team

Name	Role
Valentino Constantinou	Head of Infrastructure
John Kanalakis	Chief Technology Officer

License

This dataset is released under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

You are free to:

• Share — copy and redistribute the material in any medium or format
• Adapt — remix, transform, and build upon the material

Under the following terms:

• Attribution — You must give appropriate credit to Infactory, provide a link to the license, and indicate if changes were made.
• Non-Commercial — You may not use the material for commercial purposes without a separate commercial license from Infactory.

Commercial licensing: For commercial use, contact hello@infactory.ai.

Citation

@dataset{camera_detection_2026,
  title={Soccer Camera Detection Dataset},
  author={Constantinou, Valentino and Kanalakis, John},
  year={2026},
  publisher={Infactory},
  url={https://huggingface.co/datasets/infactory-ai/camera-detection},
  note={Per-frame camera characterization: 9-class shot scale (CineScale-aligned), 5-class camera angle (CineScale2-aligned)}
}