A deep learning based hazardous materials (HAZMAT) sign detection robot with restricted computational resources

Sharifi, Amir; Zibaei, Ahmadreza; Rezaei, Mahdi

doi:10.1016/j.mlwa.2021.100104

Cited by 12 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Motion analysis is also one of the most fundamental and challenging problems in machine vision that can be widely used in various applications, such as automatic driving, performance detection, scene perception, and robotics [30]. Recurrent all‐pairs field transforms (RAFT) [31] and FlowNet2 [25] models are utilized in our model to extract the required optical flows for the in‐car and out‐of‐car videos, respectively.…”

Section: Methodsmentioning

confidence: 99%

DIPNet: Driver intention prediction for a safe takeover transition in autonomous vehicles

Bonyani

Rahmanian

Jahangard

et al. 2023

IET Intelligent Trans Sys

View full text Add to dashboard Cite

Following the successful development of advanced driver assistance systems (ADAS), the current research directions focus on highely automated vehicles aiming at reducing human driving tasks, and extending the operational design domain, while maintaining a higher level of safety. Currently, there are high research demands in academia and industry to predict driver intention and understating driver readiness, e.g. in response to a “take‐over request” when a transition from automated driving mode to human mode is needed. A driver intention prediction system can assess the driver's readiness for a safe takeover transition. In this study, a novel deep neural network framework is developed by adopting and adapting the DenseNet, long short‐term memory, attention, FlowNet2, and RAFT models to anticipate the diver maneuver intention. Using the public “Brain4Cars” dataset, the driver maneuver intention will be predicted up to 4 s in advance, before the commencement of the driver's action. The driver intention prediction is assessed based on 1) in‐cabin 2) out‐cabin (road) and 3) both in‐out cabin video data. Utilizing K‐fold cross‐validation, the performance of the model is evaluated using accuracy, precision, recall, and F1‐score metrics. The experiments show the proposed DIPNet model outperforms the state‐of‐the‐art in the majority of the driving scenarios.

show abstract

Section: Methodsmentioning

confidence: 99%

DIPNet: Driver intention prediction for a safe takeover transition in autonomous vehicles

Bonyani

Rahmanian

Jahangard

et al. 2023

IET Intelligent Trans Sys

View full text Add to dashboard Cite

show abstract

“…The use of image recognition technologies by machine learning requires significant computing power, which is determined by the clock rate and number of processor cores, the amount of RAM and the power of the video information processing system. Thus, it will be optimal to use a PC with maximum characteristics [9][10][11].…”

Section: Materials and Methods Of Researchmentioning

confidence: 99%

Remote Visual Information System for Identification of Dangerous Substances Using Unmanned Aircrafts

Bychenko

Udovenko

Nuianzin

et al. 2022

Advances in Science and Technology

View full text Add to dashboard Cite

Development of a functional model of the process of creating a knowledge base on the recognition of objects and actions of the enemy on the basis of neural networks and fuzzy logic. The aim of the work is to develop a set of software and hardware designed for remote identification of hazardous substances by machine visual recognition of information signs of dangerous goods with the output of relevant information to the means of visual display (interface). Recommendations concerning providing UAVs with the necessary technical means to monitor the zone of emergencies are analyzed. The recommendations of the organization of radio communication between the UAV and the operator depending on the range of the UAV departure, terrain conditions etc are analyzed and given. The structural scheme of the complex of remote recognition of HC in the form of blocks, units and software and hardware is developed. As a result of the analysis of programming systems, it was found that Python programming language is the best choice to ensure the full operation of the software due to the built-in capabilities and the involvement of third-party frameworks. A database containing information on more than 3.000 HCs with detailed recommendations for emergency response is developed. The hardware and software complex for remote identification of dangerous substances by machine visual recognition of information signs of dangerous goods by UAV, consisting of unmanned aerial platform with photo-video recording means, data transmission system to ground control station, PC for processing results and related software are substantiated and developed. A test of the UAV's capabilities in recognizing danger signs with UN numbers in different lighting conditions was tested. In all cases, the HC was accurately identified. The ideas and methods proposed in this article will allow to create cheap and simple tools for rescue units of Ukraine, which deal with the consequences of emergencies related to the leakage of HCs.

show abstract

“…With the development of convolutional neural networks (CNN), a variety of deep learning based methods have been put forward for hazmat marker detection. Considering the demand in accuracy and efficiency, Sharifi et al [8] designed a deep learning based hazmat sign detection robot with restricted computational resources. Based on a cascaded network, Wang et al [9] proposed a light convolutional neural network for license plate detection and recognition with higher accuracy and lower computational cost.…”

Section: Introductionmentioning

confidence: 99%

HMD-Net: A Vehicle Hazmat Marker Detection Benchmark

Jia

Wang

Wang³

et al. 2022

Entropy

View full text Add to dashboard Cite

Vehicles carrying hazardous material (hazmat) are severe threats to the safety of highway transportation, and a model that can automatically recognize hazmat markers installed or attached on vehicles is essential for intelligent management systems. However, there is still no public dataset for benchmarking the task of hazmat marker detection. To this end, this paper releases a large-scale vehicle hazmat marker dataset named VisInt-VHM, which includes 10,000 images with a total of 20,023 hazmat markers captured under different environmental conditions from a real-world highway. Meanwhile, we provide an compact hazmat marker detection network named HMD-Net, which utilizes a revised lightweight backbone and is further compressed by channel pruning. As a consequence, the trained-model can be efficiently deployed on a resource-restricted edge device. Experimental results demonstrate that compared with some established methods such as YOLOv3, YOLOv4, their lightweight versions and popular lightweight models, HMD-Net can achieve a better trade-off between the detection accuracy and the inference speed.

show abstract

A deep learning based hazardous materials (HAZMAT) sign detection robot with restricted computational resources

Cited by 12 publications

References 8 publications

DIPNet: Driver intention prediction for a safe takeover transition in autonomous vehicles

DIPNet: Driver intention prediction for a safe takeover transition in autonomous vehicles

Remote Visual Information System for Identification of Dangerous Substances Using Unmanned Aircrafts

HMD-Net: A Vehicle Hazmat Marker Detection Benchmark

Contact Info

Product

Resources

About