Traffic Sign Recognition using Deeplearning for Autonomous Driverless Vehicles

Prakash, A Suriya; Vigneshwaran, D; Ayyalu, R Seenivasaga; Sree, S. Jayanthi

doi:10.1109/iccmc51019.2021.9418437

Cited by 16 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The improved model had a mAP of 79.77% at 29 FPS, which was extremely beneficial for identifying labels in continuously detected videos. Prakash et al [40] proposed an extended LeNet-5 CNN model, using the Gabor-based kernel followed by the normal convolutional kernel after the pooling layer. The hue and saturation value color space features had faster detection speeds and fewer impacts from illumination.…”

Section: Studies Related To the Detection Of Traffic Signsmentioning

confidence: 99%

Application of an Effective Hierarchical Deep-Learning-Based Object Detection Model Integrated with Image-Processing Techniques for Detecting Speed Limit Signs, Rockfalls, Potholes, and Car Crashes

Chung

2023

Future Internet

View full text Add to dashboard Cite

Against the backdrop of rising road traffic accident rates, measures to prevent road traffic accidents have always been a pressing issue in Taiwan. Road traffic accidents are mostly caused by speeding and roadway obstacles, especially in the form of rockfalls, potholes, and car crashes (involving damaged cars and overturned cars). To address this, it was necessary to design a real-time detection system that could detect speed limit signs, rockfalls, potholes, and car crashes, which would alert drivers to make timely decisions in the event of an emergency, thereby preventing secondary car crashes. This system would also be useful for alerting the relevant authorities, enabling a rapid response to the situation. In this study, a hierarchical deep-learning-based object detection model is proposed based on You Only Look Once v7 (YOLOv7) and mask region-based convolutional neural network (Mask R-CNN) algorithms. In the first level, YOLOv7 identifies speed limit signs and rockfalls, potholes, and car crashes. In the second level, Mask R-CNN subdivides the speed limit signs into nine categories (30, 40, 50, 60, 70, 80, 90, 100, and 110 km/h). The images used in this study consisted of screen captures of dashcam footage as well as images obtained from the Tsinghua-Tencent 100K dataset, Google Street View, and Google Images searches. During model training, we employed Gaussian noise and image rotation to simulate poor weather conditions as well as obscured, slanted, or twisted objects. Canny edge detection was used to enhance the contours of the detected objects and accentuate their features. The combined use of these image-processing techniques effectively increased the quantity and variety of images in the training set. During model testing, we evaluated the model’s performance based on its mean average precision (mAP). The experimental results showed that the mAP of our proposed model was 8.6 percentage points higher than that of the YOLOv7 model—a significant improvement in the overall accuracy of the model. In addition, we tested the model using videos showing different scenarios that had not been used in the training process, finding the model to have a rapid response time and a lower overall mean error rate. To summarize, the proposed model is a good candidate for road safety detection.

show abstract

Section: Studies Related To the Detection Of Traffic Signsmentioning

confidence: 99%

Application of an Effective Hierarchical Deep-Learning-Based Object Detection Model Integrated with Image-Processing Techniques for Detecting Speed Limit Signs, Rockfalls, Potholes, and Car Crashes

Chung

2023

Future Internet

View full text Add to dashboard Cite

show abstract

“…CNN and RNN models are the most prominent DL approaches in the field of traffic sign detection and recognition. Suriya Prakash, et al [4] extended and developed a classical LeNet-5 CNN model, which makes use of Gabor based kernel followed by a normal convolutional kernel after the pooling layer. Their proposed CNN model was evaluated using the German Traffic Sign Benchmark and gave an accuracy of nearly 98.9%.…”

Section: Deep Learning Techniquementioning

confidence: 99%

“…As an alternative to conventional machine learning schemes, deep learning-based schemes appear to be a promising option for efficient traffic sign detection [1], [2], [3]. According to recent literature works, to address the challenge of traffic sign detection and recognition, Suriya Prakash, et al [4] proposed a LeNet-5 Convolutional Neural Network (CNN) model that possessed a high detection accuracy of nearly 98.8%. Changzhen, et al [5] implemented an advanced detection method based on a deep CNN model which also achieved a satisfactory result of above 99.0% recognition precision.…”

Section: Introductionmentioning

confidence: 99%

Deep Residual Neural Network for Efficient Traffic Sign Detection

Cai

et al. 2023

Proceedings of International Conference on Artificial Life and Robotics

View full text Add to dashboard Cite

This paper established three deep residual neural network models with different architectures for traffic sign detection. Also, a new systematic analytic hierarchy process method for model performance evaluation has been proposed, which was utilized to determine the configuration of the deep learning model. In this paper, four evaluation metrics were used for analytic hierarchy process measurement, they are accuracy, stability, response time, and system capability. Based on the Tsinghua-Tencent 100K dataset, experimental results verified the feasibility of the proposed models for traffic sign detection and recognition which has training and testing accuracy of 99.03% and 98.01% respectively.

show abstract

“…In the TSR algorithm, firstly, candidate regions are detected using the colour features of the pixels in the detection step. Following this, the cascaded feedforward neural networks with random weights (FNNRW) classifiers are used for shape and content recognition [36]. The experimental results indicate that the average running time of the whole system is less than 10 ms (which would imply the real-time basis of the algorithm), with high accuracy.…”

Section: Traffic Sign Recognitionmentioning

confidence: 99%

“…Our results later corroborate with this. The work performed by Prakash et al [36] also presents a technique wherein the accuracy of the TSR algorithm may be as high as 99% by using optimum deep learning methodologies. In our case, we shall use the off-the-shelf software base as worked upon by Fan et al [35] to present the concept.…”

Section: Traffic Sign Recognitionmentioning

confidence: 99%

Switching Trackers for Effective Sensor Fusion in Advanced Driver Assistance Systems

Deo

Palade

2022

Electronics

View full text Add to dashboard Cite

Modern cars utilise Advanced Driver Assistance Systems (ADAS) in several ways. In ADAS, the use of multiple sensors to gauge the environment surrounding the ego-vehicle offers numerous advantages, as fusing information from more than one sensor helps to provide highly reliable and error-free data. The fused data is typically then fed to a tracker algorithm, which helps to reduce noise and compensate for situations when received sensor data is temporarily absent or spurious, or to counter the offhand false positives and negatives. The performances of these constituent algorithms vary vastly under different scenarios. In this paper, we focus on the variation in the performance of tracker algorithms in sensor fusion due to the alteration in external conditions in different scenarios, and on the methods for countering that variation. We introduce a sensor fusion architecture, where the tracking algorithm is spontaneously switched to achieve the utmost performance under all scenarios. By employing a Real-time Traffic Density Estimation (RTDE) technique, we may understand whether the ego-vehicle is currently in dense or sparse traffic conditions. A highly dense traffic (or congested traffic) condition would mean that external circumstances are non-linear; similarly, sparse traffic conditions would mean that the probability of linear external conditions would be higher. We also employ a Traffic Sign Recognition (TSR) algorithm, which is able to monitor for construction zones, junctions, schools, and pedestrian crossings, thereby identifying areas which have a high probability of spontaneous, on-road occurrences. Based on the results received from the RTDE and TSR algorithms, we construct a logic which switches the tracker of the fusion architecture between an Extended Kalman Filter (for linear external scenarios) and an Unscented Kalman Filter (for non-linear scenarios). This ensures that the fusion model always uses the tracker that is best suited for its current needs, thereby yielding consistent accuracy across multiple external scenarios, compared to the fusion models that employ a fixed single tracker.

show abstract

Traffic Sign Recognition using Deeplearning for Autonomous Driverless Vehicles

Cited by 16 publications

References 12 publications

Application of an Effective Hierarchical Deep-Learning-Based Object Detection Model Integrated with Image-Processing Techniques for Detecting Speed Limit Signs, Rockfalls, Potholes, and Car Crashes

Application of an Effective Hierarchical Deep-Learning-Based Object Detection Model Integrated with Image-Processing Techniques for Detecting Speed Limit Signs, Rockfalls, Potholes, and Car Crashes

Deep Residual Neural Network for Efficient Traffic Sign Detection

Switching Trackers for Effective Sensor Fusion in Advanced Driver Assistance Systems

Contact Info

Product

Resources

About