The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

Safety Helmet Detection Dataset (YOLO Format)

A curated computer vision dataset comprising 18,159 images designed for real-time safety helmet detection in urban traffic, surveillance, and industrial scenarios.

Dataset Overview

Total Images: 18,159
Format: YOLO annotation format (.txt files containing <class-id> <x_center> <y_center> <width> <height> normalized coordinates)
Input Resolution: Preprocessed and stretched to 640x640 pixels
Archive File: helmet-detection-dataset.zip

Dataset Splits

Split	Images	Ratio
Train	16,386	~90%
Validation	1,130	~6%
Test	643	~4%

Classes & Annotations

The dataset categorizes individuals into two distinct classes:

Class ID	Class Name	Description
`0`	`Helmet`	Rider or individual properly wearing a safety helmet
`1`	`No-Helmet`	Individual without a helmet (safety infraction)

Directory Structure Inside Archive

Once uncompressed, helmet-detection-dataset.zip follows the standard Ultralytics YOLO structure:

helmet-detection-dataset/
└── dataset10kadjusted/
    ├── train/
    │   ├── images/
    │   └── labels/
    ├── val/
    │   ├── images/
    │   └── labels/
    └── test/
        ├── images/
        └── labels/

`data.yaml` Configuration

path: ../helmet-detection-dataset/dataset10kadjusted  # dataset root
train: train/images
val: val/images
test: test/images

names:
  0: Helmet
  1: No-Helmet

Download & Usage

Download and extract the dataset using standard CLI utilities:

# Download archive from Hugging Face
curl -L -o helmet-detection-dataset.zip "https://huggingface.co/datasets/DeBoniz/helmet-detection-dataset/resolve/main/helmet-detection-dataset.zip"
# Extract directly into your project workspace
unzip -q helmet-detection-dataset.zip -d .

Training with YOLO11

Run training directly with the Ultralytics CLI:

yolo task=detect mode=train \
  model=yolo11s.pt \
  data=helmet-detection-dataset/data.yaml \
  epochs=100 \
  imgsz=640

Related Resources

GitHub Repository: DeBoniz/HelmetDetectionArtificialVision
Primary Model Checkpoint: Run 21 (best.pt) achieves 94.69% mAP50 and 89.61% recall.

Downloads last month: 97

Total file size:

1.47 GB