Traffic light recognition in varying illumination using deep learning and saliency map

2014 ◽  
Author(s):  
V. John ◽  
K. Yoneda ◽  
B. Qi ◽  
Z. Liu ◽  
S. Mita
Keyword(s):  
Deep Learning ◽  
Saliency Map ◽  
Traffic Light

DeLTR: A Deep Learning Based Approach to Traffic Light Recognition

2019 ◽  
pp. 604-615
Author(s):  
Yiyang Cai ◽  
Chenghua Li ◽  
Sujuan Wang ◽  
Jian Cheng
Keyword(s):  
Deep Learning ◽  
Traffic Light

scholarly journals Lightweight PVIDNet: A Priority Vehicles Detection Network Model Based on Deep Learning for Intelligent Traffic Lights

Sensors ◽  
2020 ◽  
Vol 20 (21) ◽  
pp. 6218
Author(s):  
Rodrigo Carvalho Barbosa ◽  
Muhammad Shoaib Ayub ◽  
Renata Lopes Rosa ◽  
Demóstenes Zegarra Rodríguez ◽  
Lunchakorn Wuttisittikulkij
Keyword(s):  
Deep Learning ◽  
Traffic Control ◽  
Detection System ◽  
Lightweight Design ◽  
Vehicle Detection ◽  
Activation Function ◽  
Urban Traffic ◽  
Vehicle Classification ◽  
Traffic Light ◽  
Traffic Lights

Minimizing human intervention in engines, such as traffic lights, through automatic applications and sensors has been the focus of many studies. Thus, Deep Learning (DL) algorithms have been studied for traffic signs and vehicle identification in an urban traffic context. However, there is a lack of priority vehicle classification algorithms with high accuracy, fast processing, and a lightweight solution. For filling those gaps, a vehicle detection system is proposed, which is integrated with an intelligent traffic light. Thus, this work proposes (1) a novel vehicle detection model named Priority Vehicle Image Detection Network (PVIDNet), based on YOLOV3, (2) a lightweight design strategy for the PVIDNet model using an activation function to decrease the execution time of the proposed model, (3) a traffic control algorithm based on the Brazilian Traffic Code, and (4) a database containing Brazilian vehicle images. The effectiveness of the proposed solutions were evaluated using the Simulation of Urban MObility (SUMO) tool. Results show that PVIDNet reached an accuracy higher than 0.95, and the waiting time of priority vehicles was reduced by up to 50%, demonstrating the effectiveness of the proposed solution.

scholarly journals Cognitive Compliance: Assessing Regulatory Risk in Financial Advice Documents

2020 ◽  
Vol 34 (09) ◽  
pp. 13636-13637
Author(s):  
Wanita Sherchan ◽  
Sue Ann Chen ◽  
Simon Harris ◽  
Nebula Alam ◽  
Khoi-Nguyen Tran ◽  
...  
Keyword(s):  
Machine Learning ◽  
Risk Factors ◽  
Deep Learning ◽  
Language Processing ◽  
Regulatory Compliance ◽  
Rapid Identification ◽  
Traffic Light ◽  
Risk Status ◽  
Government Regulations ◽  
Financial Advice

This paper describes Cognitive Compliance - a solution that automates the complex manual process of assessing regulatory compliance of personal financial advice. The solution uses natural language processing (NLP), machine learning and deep learning to characterise the regulatory risk status of personal financial advice documents with traffic light rating for various risk factors. This enables comprehensive coverage of the review and rapid identification of documents at high risk of non-compliance with government regulations.

Interpretability of deep convolutional neural networks on rolling bearing fault diagnosis

2021 ◽  
Author(s):  
Huixin Yang ◽  
Xiang Li ◽  
Wei Zhang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Deep Learning ◽  
Fault Diagnosis ◽  
Deep Neural Network ◽  
Rapid Development ◽  
Saliency Map ◽  
Data Driven ◽  
Experimental Investigations ◽  
Deep Convolutional Neural Networks

Abstract Despite the rapid development of deep learning-based intelligent fault diagnosis methods on rotating machinery, the data-driven approach generally remains a "black box" to researchers, and its internal mechanism has not been sufficiently understood. The weak interpretability significantly impedes further development and applications of the effective deep neural network-based methods. This paper contributes efforts to understanding the mechanical signal processing of deep learning on the fault diagnosis problems. The diagnostic knowledge learned by the deep neural network is visualized using the neuron activation maximization and the saliency map methods. The discriminative features of different machine health conditions are intuitively observed. The relationship between the data-driven methods and the well-established conventional fault diagnosis knowledge is confirmed by the experimental investigations on two datasets. The results of this study can benefit researchers on understanding the complex neural networks, and increase the reliability of the data-driven fault diagnosis model in the real engineering cases.

scholarly journals INTEGRATED SMART TRAFFIC CONTROL SYSTEM MENUJU PEKANBARU SEBAGAI SMART CITY

JURTEKSI ◽  
2020 ◽  
Vol 7 (1) ◽  
pp. 67-74
Author(s):  
Reny Medikawati Taufiq ◽  
Sunanto Sunanto ◽  
Yoze Rizki
Keyword(s):  
Deep Learning ◽  
Control System ◽  
Object Detection ◽  
Traffic Control ◽  
Integrated System ◽  
Design Solution ◽  
Traffic Light ◽  
Traffic Control System ◽  
Smart Mobility ◽  
Smart Traffic

Abstract: Pekanbaru still using conventional traffic light control system. Pekanbaru as the capital of Riau Province is predicted  udergo the  increased of urban population by 54.5% in 2025. It is important for Pekanbaru to immediately implement smart and efficient traffic management system, so that traffic congestion can be resolved quickly. This research paper provides a design solution for smart traffic light management (Smart Traffic Control System), based on object detection technology that uses deep learning to detect the number and type of vehicles. The number of vehicle is the basis for determining the green light timer automatically. The Smart Traffic Control System (STCS) is integrated with a web based geographic information system (smart map) that can display the current condition  (picture, the number of vehicle, congestion level) of each STCS location. This integrated system has been tested on a traffic light prototype, using a mini computer and a miniature vehicle. This integrated system is able to detect 9 out of 12 vehicles, and able to send data regularly to the smart map.  Keywords: deep learning; smart mobility; smart traffic control system Abstrak: Pengaturan lampu lalu lintas di Kota Pekanbaru masih dilakukan secara  konvensional. Pekanbaru sebagai ibukota Provinsi Riau diprediksikan akan mengalami peningkatan jumlah penduduk  perkotaan sebesar 54,5% pada tahun 2025. Dengan melihat predikisi ini, penting bagi kota Pekanbaru untuk segera memiliki tata kelola lalu lintas yang cerdas dan efisien agar kemacetan dapat ditanggulangi dengan cepat. Penelitian ini memberikan rancangan solusi untuk tata kelola  lampu lalu lintas cerdas (Smart Traffic Control System), berbasis teknologi object detection  yang menggunakan deep learning untuk mendeteksi jumlah dan jenis kendaraan. Jumlah kendaraan menjadi dasar penentuan timer lampu hijau secara otomatis. Smart Traffic Control System (STCS) terintegrasi dengan sistem informasi geografis berbasis web (smart map) yang secara kontinu menerima informasi kepadatan (gambar terkini, jumlah kendaraan, level kepadatan), kemudian menampilkannya diatas peta Kota Pekanbaru. Solusi sistem terintegrasi ini telah diujikan pada sebuah prototipe lampu lalu lintas, menggunakan komputer mini  dan  miniatur kendaraan. Sistem terintegrasi ini mampu mendeteksi 9 dari 12 kendaraan, dan mampu mengirimkan data secara berkala kepada smart map. Kata kunci: deep learning; smart mobility; smart traffic control system

scholarly journals Deep learning saliency maps do not accurately highlight diagnostically relevant regions for medical image interpretation

2021 ◽  
Author(s):  
Adriel Saporta ◽  
Xiaotong Gui ◽  
Ashwin Agrawal ◽  
Anuj Pareek ◽  
Steven QH Truong ◽  
...  
Keyword(s):  
Deep Learning ◽  
Medical Image ◽  
Image Interpretation ◽  
Saliency Map ◽  
X Rays ◽  
Clinical Practices ◽  
Influence Model ◽  
Saliency Maps ◽  
Generating Method ◽  
Low Performance

AbstractDeep learning has enabled automated medical image interpretation at a level often surpassing that of practicing medical experts. However, many clinical practices have cited a lack of model interpretability as reason to delay the use of “black-box” deep neural networks in clinical workflows. Saliency maps, which “explain” a model’s decision by producing heat maps that highlight the areas of the medical image that influence model prediction, are often presented to clinicians as an aid in diagnostic decision-making. In this work, we demonstrate that the most commonly used saliency map generating method, Grad-CAM, results in low performance for 10 pathologies on chest X-rays. We examined under what clinical conditions saliency maps might be more dangerous to use compared to human experts, and found that Grad-CAM performs worse for pathologies that had multiple instances, were smaller in size, and had shapes that were more complex. Moreover, we showed that model confidence was positively correlated with Grad-CAM localization performance, suggesting that saliency maps were safer for clinicians to use as a decision aid when the model had made a positive prediction with high confidence. Our work demonstrates that several important limitations of interpretability techniques for medical imaging must be addressed before use in clinical workflows.

Sign in / Sign up

Export Citation Format

Share Document