infactory-ai/soccer-events

Soccer (Football) In-Match Events

A curated sample dataset of in-match soccer/football events, automatically identified using multi-modal models from Infactory. This dataset sample contains video clips and metadata for three key event types: yellow cards, red cards, and goals. Infactory models are capable of detecting corner kicks, blocked shots, saved shots, substitutions, and many more types of events in gameplay footage.

Dataset Description

This dataset provides 211 annotated video clips extracted from professional soccer broadcasts, totaling approximately 3.5 hours (12,677 seconds) of footage and 190,000 frames. Each sample includes:

• A video clip (MP4, H.264/AVC, 1280×720 @ 15 FPS)
• Rich metadata including teams, competition, and precise timing

Average clip duration is 60 seconds, making this suitable for video classification and temporal action recognition tasks.

Preview

Sample event frames: Goal | Yellow Card | Red Card

Event Categories

Category	Description	Samples
yellow_card	Referee shows yellow card to a player	100
red_card	Referee shows red card to a player	11
goal	Goal being scored	100

Source Data

• Competition: Serie A (Italian top-flight football)
• Season: 2024-25
• Provider: Infront Sports & Media
• Broadcast Quality: HD 720p (1280x720)

Dataset Structure

infactory-ai/soccer-events/
├── README.md                    # Dataset card
├── metadata.csv                 # Index with all metadata
├── dataset_info.json            # Dataset statistics
└── data/
    ├── {uuid}.json              # Event metadata (211 files)
    └── {uuid}.mp4               # Video clips (211 files)

Metadata Fields

Field	Type	Description
asset_id	string	Unique identifier (UUID)
event_type	string	Original event type name
event_category	string	Normalized: yellow_card, red_card, goal
event_subtype	string	Sub-category (e.g., "Header", "Penalty")
team	string	Team involved: "Home" or "Away"
home_team	string	Home team name
away_team	string	Away team name
competition	string	Competition name
game_date	string	Match date (YYYY-MM-DD)
season	string	Season (e.g., "2024-25")
starttimecode	string	Clip start (HH:MM:SS:FF)
endtimecode	string	Clip end (HH:MM:SS:FF)
frame_rate	string	Video frame rate
duration_seconds	float	Clip duration in seconds

Usage

Loading Metadata

import pandas as pd

df = pd.read_csv("hf://datasets/infactory-ai/soccer-events/metadata.csv")

# Filter by event type
goals = df[df["event_category"] == "goal"]
red_cards = df[df["event_category"] == "red_card"]
yellow_cards = df[df["event_category"] == "yellow_card"]

print(f"Goals: {len(goals)}, Red Cards: {len(red_cards)}, Yellow Cards: {len(yellow_cards)}")

Downloading and Playing Videos

from huggingface_hub import hf_hub_download
import cv2

# Download a specific video
video_path = hf_hub_download(
    repo_id="infactory-ai/soccer-events",
    filename="data/0293e19b-7281-4f97-b2e9-57b3cee260b2.mp4",
    repo_type="dataset"
)

# Load with OpenCV
cap = cv2.VideoCapture(video_path)
fps = cap.get(cv2.CAP_PROP_FPS)
frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
print(f"Video: {fps} FPS, {frame_count} frames, {frame_count/fps:.1f}s duration")

# Read frames
while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break
    # Process frame...
cap.release()

Extracting Frames for Training

from huggingface_hub import hf_hub_download
import cv2
import pandas as pd

# Load metadata
df = pd.read_csv("hf://datasets/infactory-ai/soccer-events/metadata.csv")

# Download and extract frames from a goal video
row = df[df["event_category"] == "goal"].iloc[0]
video_path = hf_hub_download(
    repo_id="infactory-ai/soccer-events",
    filename=f"data/{row['mp4_file']}",
    repo_type="dataset"
)

cap = cv2.VideoCapture(video_path)
frames = []
while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break
    frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    frames.append(frame_rgb)
cap.release()

print(f"Extracted {len(frames)} frames from {row['event_type']} event")

Using with PyTorch DataLoader

import torch
from torch.utils.data import Dataset, DataLoader
from huggingface_hub import hf_hub_download
import pandas as pd
import cv2

class SoccerEventsDataset(Dataset):
    def __init__(self, event_category=None):
        self.df = pd.read_csv("hf://datasets/infactory-ai/soccer-events/metadata.csv")
        if event_category:
            self.df = self.df[self.df["event_category"] == event_category]
    
    def __len__(self):
        return len(self.df)
    
    def __getitem__(self, idx):
        row = self.df.iloc[idx]
        video_path = hf_hub_download(
            repo_id="infactory-ai/soccer-events",
            filename=f"data/{row['mp4_file']}",
            repo_type="dataset"
        )
        # Load video frames as tensor...
        return {"video_path": video_path, "label": row["event_category"]}

# Create dataloader for goals only
dataset = SoccerEventsDataset(event_category="goal")
loader = DataLoader(dataset, batch_size=4, shuffle=True)

License

This dataset contains proprietary content from Infront Sports & Media. Usage is restricted to non-commercial research purposes. For commercial and large-scale needs, contact Infactory.

Citation

@dataset{soccer_events_2026,
  title={Soccer (Football) In-Match Events},
  author={John Kanalakis, Valentino Constantinou},
  year={2026},
  publisher={Infactory},
  url={https://huggingface.co/datasets/infactory-ai/soccer-events}
}