Traffic Congestion Detection: Learning from CCTV Monitoring Images using Convolutional Neural Network

Kurniawan, Jason; Syahra, Sensa G. S.; Dewa, Chandra Kusuma; Afiahayati, FNU

doi:10.1016/j.procs.2018.10.530

Cited by 49 publications

(21 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The deep learning models described in Section 3.4 were implemented and re-trained on the Trafficdb video dataset in a 10-fold cross validation setup. The ResNet [51] and the deep network architecture proposed in [52] were originally tested by respective authors on a two class (Heavy vs. Light) traffic state classification. To perform similar tests, at a first stage, samples labeled as ‘Medium’ were removed from the Trafficdb: results are shown in Table 4.…”

Section: Experiments and Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Vehicular Traffic Congestion Classification by Visual Features and Deep Learning Approaches: A Comparison

Impedovo

Balducci

Dentamaro

et al. 2019

Sensors

View full text Add to dashboard Cite

Automatic traffic flow classification is useful to reveal road congestions and accidents. Nowadays, roads and highways are equipped with a huge amount of surveillance cameras, which can be used for real-time vehicle identification, and thus providing traffic flow estimation. This research provides a comparative analysis of state-of-the-art object detectors, visual features, and classification models useful to implement traffic state estimations. More specifically, three different object detectors are compared to identify vehicles. Four machine learning techniques are successively employed to explore five visual features for classification aims. These classic machine learning approaches are compared with the deep learning techniques. This research demonstrates that, when methods and resources are properly implemented and tested, results are very encouraging for both methods, but the deep learning method is the most accurately performing one reaching an accuracy of 99.9% for binary traffic state classification and 98.6% for multiclass classification.

show abstract

Section: Experiments and Discussionmentioning

confidence: 99%

“…The output is a binary classification (congested/not congested). The ResNET has been also re-trained in [52] on the Shaanxi Province dataset.…”

Section: Methodsmentioning

confidence: 99%

Vehicular Traffic Congestion Classification by Visual Features and Deep Learning Approaches: A Comparison

Impedovo

Balducci

Dentamaro

et al. 2019

Sensors

View full text Add to dashboard Cite

show abstract

“…loop detectors) are able to measure the traffic volume and speed. However, measuring the density is difficult which have recently attracted more attention (Chung & Sohn, 2018;Kurniawan et al, 2018). Normally, the traffic densities are able to classify different conditions following the time of a day.…”

Section: Image-based Traffic Density Classificationmentioning

confidence: 99%

An image generation approach for traffic density classification at large-scale road network

Cho

Jung

et al. 2020

Journal of Information and Telecommunication

View full text Add to dashboard Cite

Recently, with the rapid development of deep learning models, traffic analysis using image datasets recently has attracted more attention. Specifically, the network traffic can be represented to images as the input for deep learning models to provide various applications (e.g. Spatio-Temporal traffic forecasting). In this study, we propose a new image generation approach for traffic density classification in terms of large-scale road network. Particularly, traffic volume and speed are at certain areas able to be measured by using surveillance systems (e.g. loop detectors). However, measuring the density is difficult which depends on the spatial correlation from the perspective of the network. Consequently, an effective image generation approach, based on information arrival and departure time of vehicles, is proposed to deal with this problem. Regarding the experiment, traffic density classification using a convolutional neural network is executed on roadside equipment data of 11 continuous intersections for evaluating the effectiveness of the proposed approach.

show abstract

“…Riset berikut menggunakan deep learning untuk membangun sistem peringatan darurat dari data video CCTV (Kang & Choo, 2016). Riset lain adalah mendeteksi kemacetan lalu-lintas dari video CCTV dengan menggunakan Convolutional Neural Network (Kurniawan, Dewa, & Afiahayati, 2018).…”

Section: Pendahuluanunclassified

Object Counting Pada Data Video

Herteno¹,

Faisal²,

Nugroho³

et al. 2020

klik

View full text Add to dashboard Cite

One object counting implementation is counting the number of road users from video data sources obtained from CCTV streaming. Video processing on CCTV is usually done on the server side by sending video data. If the need is only to determine the density of traffic, then the method is considered too expensive to be implemented because of the cost of internet connection and bandwidth that must be spent. The solution is to use a small computing device that can process the video first, and the calculation results are sent to the server regularly. In this study, a comparison between the Tensorflow Object Counting learning algorithm and the MOG2 Background Subtractor image processing algorithm with the aim to determine the accuracy of the calculation. The result is known that better accuracy is given by the MOG2 Background Subtractor technique and also the process is carried out using only a small percentage of the amount of memory and processor compared to the Tensorflow Object Counting technique. MOG2 Background Substractor technique is expected to be used on devices that have small data sourcesKeywords : Object Counting, Tensorflow, MOG2 Background SubstractorSalah satu implementasi object counting adalah menghitung jumlah pengguna jalan dari sumber data video yang didapat dari streaming CCTV. Pemprosesan video pada CCTV biasanya dilakukan disisi server dengan mengirimkan data video. Jika keperluannya hanya untuk mengetahui kepadatan lalu lintas, maka cara tersebut dinilai terlalu mahal untuk diimplementasikan karena biaya koneksi internet dan bandwidth yang harus dikeluarkan. Pemecahannya adalah menggunakan perangkat komputasi kecil yang dapat memproses video tersebut terlebih dahulu, dan hasil perhitungannya dikirimkan ke server secara berkala. Pada penelitian ini dilakukan perbandingan antara algoritma pembelajaran Tensorflow Object Counting dan algoritma image processing MOG2 Background Substractor dengan tujuan untuk mengetahui akurasi penghitungan. Hasilnya diketahui akurasi yang lebih baik diberikan oleh teknik MOG2 Background Substractor dan juga proses yang dilakukan hanya menggunakan prosentase jumlah memori dan prosessor yang kecil dibandingkan teknik Tensorflow Object Counting. Sehingga teknik MOG2 Background Substractor ini diharapkan dapat digunakan pada perangkat yang memiliki sumber data kecil. Kata kunci : Object Counting, Tensorflow, MOG2 Background Substractor.

show abstract

Traffic Congestion Detection: Learning from CCTV Monitoring Images using Convolutional Neural Network

Cited by 49 publications

References 9 publications

Vehicular Traffic Congestion Classification by Visual Features and Deep Learning Approaches: A Comparison

Vehicular Traffic Congestion Classification by Visual Features and Deep Learning Approaches: A Comparison

An image generation approach for traffic density classification at large-scale road network

Object Counting Pada Data Video

Contact Info

Product

Resources

About