Reliable and Rapid Traffic Congestion Detection Approach Based on Deep Residual Learning and Motion Trajectories

Abstract

Traffic congestion detection systems help manage traffic in crowded cities by analyzing videos of vehicles. Existing systems largely depend on texture and motion features. Such systems face several challenges, including illumination changes caused by variations in weather conditions, complexity of scenes, vehicle occlusion, and the ambiguity of stopped vehicles. To overcome these issues, this paper proposes a rapid and reliable traffic congestion detection method based on the modeling of video dynamics using deep residual learning and motion trajectories. The proposed method efficiently uses both motion and deep texture features to overcome the limitations of existing methods. Unlike other methods that simply extract texture features from a single frame, we use an efficient representation learning method to capture the latent structures in traffic videos by modeling the evolution of texture features. This representation yields a noticeable improvement in detection results under various weather conditions. Regarding motion features, we propose an algorithm to distinguish stopped vehicles and background objects, whereas most existing motion-based approaches fail to address this issue. Both types of obtained features are used to construct an ensemble classification model based on the support vector machine algorithm. Two benchmark datasets are considered to demonstrate the robustness of the proposed method: the UCSD dataset and NU1 video dataset. The proposed method achieves competitive results (97.64% accuracy) when compared to state-ofthe-art methods.

Sustainable Traffic Regulation System in Protected Areas: Pilot Technology Testing in National Park in the Czech Republic

“…As part of the quality verification of the license plate recognition subsystem, it was determined that the installed fixed point cameras show high reliability of almost above 95% when using two cameras and 90% when using one camera. The reliability values are absolutely sufficient for systems of similar type, as also evidenced by publications focusing on tests by other authors [22][23][24][25]. In general, it is advisable to place the cameras with license plate detection as close as possible to the communication.…”

A Frame Work Designing for Deep Fake Motion Detection using Deep Learning in Video Surveillance Systems

“…Nguyen et al [5] proposed a face recognition method that combines a learning framework. M. Hori et al [6] identified a major challenge of facial recognition as the large appearance variations caused by factors such as viewpoints.…”

Robust Traffic Congestion Recognition in Videos Based on Deep Multi-Stream LSTM

“…However, depending solely on one feature has its shortcoming. In our previous work [17], a congestion classification was achieved by aggregating the output from separate motion-based and texture-based classifiers. Motion features were generated by averaging motion trajectories every batch of frames, while the texture features were obtained by generating compact texture vectors using the learning-to-rank technique [18].…”