A Novel Approach of Traffic Density Estimation Using CNNs and Computer Vision

Abstract

In modern life, we face many problems, one of which is the increasingly serious traffic jam. The cause is the large volume of vehicles, inadequate infrastructure and unreasonable distribution, and ineffective traffic signal control. This requires finding methods to optimize traffic flow, especially during peak hours. To optimize traffic flow, it is necessary to determine the traffic density at each time in the streets and intersections. This paper proposed a novel approach to traffic density estimation using Convolutional Neural Networks (CNNs) and computer vision. The experimental results with UCSD traffic dataset show that the proposed solution achieved the worst estimation rate of 98.48% and the best estimation rate of 99.01%.

Deep Learning based Artificial Intelligent Systems in Road Traffic Density Estimation and Congestion Classification

“…From the Table 2 above, we say that the proposed deep method performs well compared to other three state of the art methods (9,(22)(23)(24) and produces exemplary output results.…”

“…Hence, a comparative analysis of the proposed DCNN with other existing methods on the publicly available UCSD Trafficdb dataset is made and is listed in below Table 2. The below table presents a table of comparison with the other existing deep learning methodologies, J Kurniawan, et al (9) , Sabbani Imad, et al (22) , D Impedovo, et al (23) and Nguyen L.A, et al(2021) (24) in the literature towards density estimation and congestion classification techniques.…”

“…Here, in the proposed system, a simpler dcnn architecture provides more interpretability and insight into the learned representation and features and performs better than Resnet for the road traffic density estimation and congestion classification task. The work (24) has a preprocess model where the similar frames are cut an discarded and then passed onto a segmentation model where the non vehicle objects are segmented and discarded. The output of the segmentation module is then fed into cnn model.…”

“…The output of the segmentation module is then fed into cnn model. The proposed model is superior to (24) in a way where the raw road vehicle images are fed into deep cnn and produces a remarkable test result.…”