Thought Leadership

Helmet detection in 2-wheeler

24 Apr, 2023


Wearing Helmet is a critical safety measure not only for riders but also for passengers. However, people often tend to skip wearing these protective headgears, thereby leading to, an increased risk of injury or death in the event of an accident. There is a growing necessity to develop innovative methods that automatically monitor and prevent unsafe driving. To address this issue, we have developed a computer vision-based helmet detection system that can detect if a rider has his helmet on in real-time.

This paper presents a comprehensive review of the state-of-the-art computer vision-based helmet detection systems for two-wheelers. The review covers various aspects of helmet detection, including image pre-processing, feature extraction, and classification. The strengths and limitations of existing approaches are analyzed, while proposing potential directions for future research. The results demonstrate that computer vision-based helmet detection systems hold significant potential to reduce the risk of accidents and improve safety for riders.


According to a report released by the Ministry of Road Transport and Highways, a total of 46,593 individuals who were not wearing helmets lost their lives in road accidents in the year 2021. The report indicates that out of these, 32,877 were drivers, while 13,716 were passengers. It further highlights that the severity of accidents, measured by the number of deaths per 100 accidents, has been increasing since 2000. The report emphasizes the need for better traffic management strategies that can help reduce the impact of crashes and prevent tragic losses of life on the roads. In this paper, we aim to automatically detect the riding helmets (e.g., whether the rider wears the helmet or not) by analyzing the riding surveillance images with and without helmets. Based on the collected images, we first detect the object of interest (i.e., Rider) and further analyze whether the rider wears the helmet or not, by using computer vision and machine learning techniques.

System Overview

The system consists of a high-resolution camera embedded in a 2-wheeler instrument cluster to scan the rider’s head and use computer vision algorithm to identify whether the rider is wearing helmet or not. Instrument cluster will send ignition on command to engine ECU only if the rider is seen wearing helmet. For object detection, we incorporate frequency domain information of the image with a popular human detection algorithm Histogram of Oriented Gradient (HOG) for rider detection; in the second step, the combination of color-based and Circle Hough Transform (CHT) feature extraction techniques is applied to detect different helmets used for riding.

Working principle of Helmet detection system

The working principle of helmet detection technology is based on computer vision and deep learning algorithms. The camera mounted on the instrument cluster captures images of the rider. These images are then processed using computer vision algorithms to detect the presence or absence of a helmet on the rider’s head. The deep learning algorithms are trained using thousands of images of riders wearing and not wearing helmets. This training data helps the system learn to accurately detect helmets under various lighting conditions, several types of helmets, and even when riders are wearing sunglasses or other accessories.

Helmet detection using computer vision

Detecting helmets using computer vision involves several engineering techniques, including object detection, feature extraction, and classification. These techniques can be combined and optimized to create an effective and reliable helmet detection system that can improve rider safety on the road.

Object detection algorithms use a variety of mathematical and statistical methods to extract features from the image and identify the objects of interest. Standard object detection algorithms include Haar Cascade, Histogram of Oriented Gradients (HOG), or deep learning-based methods such as Convolutional Neural Networks (CNNs). We utilize Haar Cascade classifier in our paper, which uses a set of rectangular Haar-like features to detect objects.

The Haar-like features can be represented mathematically as follows:

f(x)= sum(w(i)* p(i)), i = 1 to n

where f(x) is the feature function, w(i) is the weight of the i-th rectangle, and p(i ) Is the intensity value of the i-th rectangle. The weights and intensity values are learned from a training set of images and are used to identify the object in new images

Once the rider has been identified, the computer vision system can then analyze the image to determine whether they are wearing a helmet. This involves identifying the features of the helmet, such as its shape, size, and color. These features can be extracted using various feature extraction techniques such as Scale Invariant Feature Transform (SIFT), Speeded Up Robust Features (SURF), or Local Binary Patterns (LBP).

For example, the SIFT algorithm works by identifying KeyPoint in the image and extracting local feature descriptors around each KeyPoint. The local feature descriptors are represented as a set of vectors that capture the orientation, scale, and intensity of the local image patch. The feature vectors can be mathematically represented as follows:

V = (v (1), v (2), …, v(n))

where v(i) is the i-th element of the feature vector and n is the number of dimensions of the feature space.

Finally, a classification algorithm is used to determine whether the rider is wearing a helmet or not based on the extracted features. This can be done using machine learning techniques such as Support Vector Machines (SVMs), Random Forests, or Deep Neural Networks (DNNs).

The SVM can be mathematically represented as follows:

f(x)= sign(sum(alpha(i)* y(i)* K(x, x(i)))- b)

where f(x) is the classification function, alpha(i ) Is the weight assigned to the i-th training example, y(i) is the class label of the i-th training example, K (x, x(i)) is the kernel function that measures the similarity between two feature vectors, and b is the bias term. The SVM learns the optimal values of the weight vector and the bias term during training and uses them to classify new images.

Once the computer vision system has detected that the rider is not wearing a helmet, it can trigger an alert or safety system to remind the rider to wear a helmet, or even automatically shut off the engine until the rider put on their helmet.


Helmet detection technology can help improve road safety and reduce the number of accidents and fatalitiesdue to non-use of helmets. The technology integrated to existing security system provides realtime aler to the riders to wear their helmet immediately before they continue their journey. Secondly, it can help in enforcing helmet laws more effectively. Often, traffic police are unable to put surveillance on motorist notwearing helmets due to the sheer volume of traffic on the roads. Helmet detection technology can help automate the process and make it easier to enforce the law.

Conclusion and future work

In conclusion, the proposed computer vision-based system for detecting helmet use in motorcycle riders holds enormous potential for improving road safety. By integrating with the engine ECU, the system can ensure that the motorcycle is only started if the rider is wearing a helmet. Future research can focus on improving the real-time performance and robustness of the system and integrating it with other safety systems. Additionally, there is a need to develop systems that can detect helmets for multiple riders and other types of protective gear. Addressing privacy concerns, augmenting data sets, exploring edge computing, and addressing ethical considerations are also key areas for future research. By improving the accuracy and reliability of these systems, we can enhance rider safety and reduce the risk of accidents.

Subscribe to our newsletter