Real-time detection and classification of traffic light signals

Said, Asaad F.; Hazrati, Mehrnaz Kh.; Akhbari, Farshad

doi:10.1109/aipr.2016.8010547

Cited by 6 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While, [17], [18] and [19] used HSI color space. A CIELab color model is used in research works [20], [21] and [22]. Color spaces YCbCr, YUV and HCL are also presented in research works [23], [24] and [25] respectively.…”

Section: State Of the Artmentioning

confidence: 99%

See 1 more Smart Citation

Traffic Light Detection: A cost effective approach

Iftikhar,

Riaz,

Ali

et al. 2021

VFAST trans. softw. eng.

View full text Add to dashboard Cite

In last couple of decades, the technological advancements in image and video processing has brought great revolution in our life. Some of the key areas where these advancements have played a key role are: autonomous vehicles, drone technology, crowd monitoring, traffic monitoring, object tracking etc. Nowadays a lot of work is under process for improving capabilities of autonomous vehicles and driver assisted systems. Our focus in this paper is related to automated traffic light detection system with improved capabilities in terms of time complexity and accuracy. The time complexity is directly related to image or video quality with regard to resolution of video and the accuracy is often compromised because of identification of similar objects. The similar objects often appear in video frames when each frame of video is analyzed completely. In order to solve the problem of real time detection of traffic lights in a high-resolution video having 30 frames per second with a resolution of 1280 × 720, we propose an algorithm that systematically searches in middle 70% region of each frame. The proposed algorithm optimizes the search space by dividing middle region into 3. There are three methods for searching and registering a traffic light is proposed in this paper. The basic concept is at single instance a traffic light can exist on one of these three regions. These trategies help in reducing computation complexity tremendously. The Hough Circle Transform technique of image processing is exploited to accurately detect red and green circles of light in the traffic light. Efficacy of the proposed technique in terms of improved time and accuracy is demonstrated on a real dataset collected from Nexar (dashcam mobile solution provider), it encompasses different illumination conditions: day, evening, night, cludy weather and rain etc.

show abstract

Section: State Of the Artmentioning

confidence: 99%

“…Noise in images due to distortion, camera vibration and similar objects can affect the traffic light detection. Morphological operations are performed on binary images including Dilation, Erosion and Filling to remove the noise [22]. Dilation and Erosion are widely used in literature.…”

Section: State Of the Artmentioning

confidence: 99%

Traffic Light Detection: A cost effective approach

Iftikhar,

Riaz,

Ali

et al. 2021

VFAST trans. softw. eng.

View full text Add to dashboard Cite

show abstract

“…Yu Tiancheng et al proposed an improved deep learning method based on the Faster RCNN model [1]; Liu Keqi et al proposed a traffic signal recognition method based on image enhancement and verified the method under complex background conditions [2]. Wang Ying et al proposed a traffic light recognition and classification method using the YOLO model based on deep learning and proved that the model has high accuracy and real-time performance [3]; Zhou Xuanru et al proposed a real-time traffic signal recognition algorithm based on HOG feature and SVM [4]; Behrendt et al proposed a complete system of traffic signal detector, tracker and classifier based on deep learning, stereo vision and vehicle odometer, Real-time perception of traffic lights [5]; Weber et al proposed a camera-based realtime detection and classification system for traffic lights [6]; Said et al proposed a powerful real-time method to detect traffic lights and recognize their status in complex traffic scenes based only on image processing techniques [7].…”

Section: Introductionmentioning

confidence: 99%

Research on application of meta-learning in traffic light recognition

Wang

Yang

et al. 2023

Third International Conference on Digital Signal and Computer Communications (DSCC 2023)

View full text Add to dashboard Cite

Traffic light recognition is essential to intelligent vehicle information perception, and its accuracy directly affects traffic safety. Based on sorting out and analyzing the existing related research on traditional machine learning and deep learning in traffic light recognition, the basic principles of two standard algorithms in meta-learning are introduced in detail. Compared with traditional machine learning, using the MAML algorithm and Reptile algorithms in meta-learning to recognize traffic lights by selecting the WPI dataset and programming in PYTHON. The simulation results show that the recognition accuracy based on the meta-learning algorithm is higher than that based on the traditional machine learning algorithm; In the meta-learning algorithm, the Reptile algorithm outperforms the MAML algorithm.

show abstract

“…Within the candidate regions, the features known by humans were selected to detect the traffic light early including shape [5], colour [6], structure [10], and area and aspect ratio [11]. To overcome the limitations of human knowledge, some intelligent classifiers were proposed thereafter.…”

Section: Introductionmentioning

confidence: 99%