Traffic Sign Recognition With Lightweight Two-Stage Model in Complex Scenes

Wang, Zhengshuai; Wang, Jianqiang; Li, Yali; Wang, Shengjin

doi:10.1109/tits.2020.3020556

Cited by 23 publications

(12 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Their fastest combination can reach 7.3 FPS on a GPU-supported Snapdragon TM 820A processor suited for neural processing. Wang and colleagues [40] proposed to restrict the search space by exploring the distribution of location and scales of traffic signs, and then a two-stage detector based on R-FCN [41]: a superclass detector is applied first, and a refinement stage identifies individual traffic sign classes. However, the previous knowledge on scale/locations is highly dependent on camera parameters (intrinsic and pose) and the training set.…”

Section: Related Workmentioning

confidence: 99%

“…As in [33,40], our paper also explores image crops to accelerate the TSR task, but in a much more systematic manner. Given a camera with known intrinsic parameters, we use an online extrinsic calibration scheme [42] and select Regions of Interest (ROIs) where a traffic sign is expected to be, as well as the expected scale of the sign at each ROI.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Lightweight Traffic Sign Detection using Detachable Onboard Cameras

Paula

Jung

2023

Preprint

View full text Add to dashboard Cite

This paper presents a novel approach for traffic sign recognition (TSR) using off-the-shelf onboard vehicular cameras. Assuming that the camera intrinsic parameters are obtained offline, an online calibration scheme is used to estimate the extrinsic camera parameters, and Regions of Interest (ROIs) are created in the image domain based on the expected geometry and location of the traffic signs. Within theseROIs, the scale variation of the sign and background complexity are limited, allowing the development of lightweight Convolutional Neural Networks (CNNs) for TSR. Our experimental results for Brazilian traffic signs indicate that the proposed approach presents better accuracy than state-of-the-art methods at faster running times.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Lightweight Traffic Sign Detection using Detachable Onboard Cameras

Paula

Jung

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Важливою частиною інтелектуальної транспортної системи є розпізнавання дорожніх знаків з високою точністю, та в режимі реального часу. В [11] пропонується новий та гнучкий двоетапний підхід, що базується на великомасштабних дорожніх знаках і внутрішньому конфлікті між регресією розташування та класифікацією дорожніх знаків. Він поєднує в собі легкий детектор суперкласу з уточненим класифікатором.…”

Section: вступunclassified

Comparative analysis of the driver's psychological perception of information and the use of road sign recognition systems

Barvinska¹,

Hrytsun²

2023

JMET

View full text Add to dashboard Cite

This article analyzes drivers' psychophysiological perception of information on the road and the advantages of using means of automatic use of signs (TSR). A survey of drivers was conducted on the road section where traffic organization changed. The drivers were chosen with different driving experiences, age categories, and needs for using the car, but they used the road section under investigation even before its reconstruction. Drivers of vehicles by age category were divided into three categories: under 25 years of age (category 1), 42% of drivers aged 26 to 50 years (category 2), and 19% of drivers aged 50 and older (category 3). It was established that 47% of the first drivers' category use automatic road sign recognition tools, 31% of the second category use the TSR system, and only 22% of the third category use the road sign recognition system. Four new road signs were installed during the development of the design schemes for organizing traffic in the middle section at a distance of 50 m. Based on this, an additional survey was conducted on drivers' memorization of specific new signs installed on the investigated section of the road. The results of the survey of drivers of different age categories were taken into account. It was studied that the most perceived number of road signs for the third category of drivers are observed at a distance of 50 to 150 m. At a distance of 50 to 150 m, they concentrate their attention, and after 150 m, they forget about the changed scheme in the traffic organization. In conclusion, drivers, getting used to traffic routes, lose vigilance, and pay less attention to existing information signals, which causes them to make wrong decisions when changing traffic organization on certain road sections. It is proposed to use automatic road sign recognition tools that are not affected by external and internal factors to increase the reliability of drivers and ensure road safety.

show abstract

“…Only 3% of error rate is reported by using the proposed approach on the proposed dataset. Wang et al in [29] presented a two-stage network model for the traffic sign detection. The authors have used the prior information such as locations and sizes of the traffic signs to make a probability distribution model.…”

Section: Related Workmentioning

confidence: 99%

Small Traffic Sign Detection in Big Images: Searching Needle in a Hay

et al. 2022

View full text Add to dashboard Cite

Traffic sign detection is an essential module of self-driving cars and driver assistance system. The major challenge being, traffic sign appear relatively smaller in road view images. It covers only 1%-2% of the total image area. Hence, its challenging to detect very small traffic sign in a larger image covering huge background of similar shape objects. Thus, we propose YOLOv3 network layers pruning and patch wise training strategy for small sized traffic sign detection. This aids in improving recall percentage and mean Average Precision. We also propose anchor box selection algorithm that uses bounding box dimension density to obtain optimal anchor set for the dataset. This reduces false positives and log-average miss rate. The proposed approach is evaluated on German traffic sign detection benchmark and Swedish traffic sign dataset and proves that it achieved a good balance between mAP and inference time.

show abstract

Traffic Sign Recognition With Lightweight Two-Stage Model in Complex Scenes

Cited by 23 publications

References 35 publications

Lightweight Traffic Sign Detection using Detachable Onboard Cameras

Lightweight Traffic Sign Detection using Detachable Onboard Cameras

Comparative analysis of the driver's psychological perception of information and the use of road sign recognition systems

Small Traffic Sign Detection in Big Images: Searching Needle in a Hay

Contact Info

Product

Resources

About