License plate recognition using neural architecture search for edge devices

Shashirangana, Jithmi; Padmasiri, Heshan; Meedeniya, Dulani; Perera, Charith; Nayak, Soumya Ranjan; Nayak, Janmenjoy; Vimal, S.; Kadry, Seifedine

doi:10.1002/int.22471

Cited by 22 publications

(11 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Edge computing supports solves this issue by merely increasing the computational capabilities of the edge devices, thus reducing the communication cost and the application latency. Moreover, it has become possible to due to the increase in computational performance in edge devices without significantly compromising energy efficiency [57].…”

Section: Alpr Using Edge Devicesmentioning

confidence: 99%

“…In our previous study [57], we have discussed the architecture of the Lite LP-Net models in detail. As the next phase, this paper mainly describes the hardware circuit configurations from the deployment point of view, synthetic data generation process, stochastic super-network implementation and the bi-level optimization in Section 3, as the scientific contribution.…”

Section: Alpr Using Edge Devicesmentioning

confidence: 99%

“…Our previous work has presented the detailed implementation aspects of the LP-net architecture used for this study [57].…”

Section: Search Strategymentioning

confidence: 99%

See 2 more Smart Citations

Automated License Plate Recognition for Resource-Constrained Environments

Padmasiri

Shashirangana

Meedeniya

et al. 2022

Sensors

Self Cite

View full text Add to dashboard Cite

The incorporation of deep-learning techniques in embedded systems has enhanced the capabilities of edge computing to a great extent. However, most of these solutions rely on high-end hardware and often require a high processing capacity, which cannot be achieved with resource-constrained edge computing. This study presents a novel approach and a proof of concept for a hardware-efficient automated license plate recognition system for a constrained environment with limited resources. The proposed solution is purely implemented for low-resource edge devices and performed well for extreme illumination changes such as day and nighttime. The generalisability of the proposed models has been achieved using a novel set of neural networks for different hardware configurations based on the computational capabilities and low cost. The accuracy, energy efficiency, communication, and computational latency of the proposed models are validated using different license plate datasets in the daytime and nighttime and in real time. Meanwhile, the results obtained from the proposed study have shown competitive performance to the state-of-the-art server-grade hardware solutions as well.

show abstract

Section: Alpr Using Edge Devicesmentioning

confidence: 99%

Section: Alpr Using Edge Devicesmentioning

confidence: 99%

See 1 more Smart Citation

Automated License Plate Recognition for Resource-Constrained Environments

Padmasiri

Shashirangana

Meedeniya

et al. 2022

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Another method has been to use computer vision techniques to detect licence plate numbers. There are many challenges [9] to detecting vehicle licence plates covertly in remote jungle areas, such as limited visibility and lighting [10], limited connectivity, limited access to computation and energy sources [11], exposure to harsh jungle environments (e.g. humidity, water etc.)…”

Section: Vehicle Tracking Using Rfid/ble/vision Technologymentioning

confidence: 99%

Low-Cost SMS Driven Location Tracking Platform Towards Anti-Poaching Efforts

Burkett¹,

Wengel²,

Goossens³

et al. 2022

Preprint

View full text Add to dashboard Cite

Throughout the world, poaching has been an ever-present threat to a vast array of species for over many decades. Traditional anti-poaching initiatives target catching the poachers. However, the challenge is far more complicated than catching individual poachers. Poaching is an industry which needs to be fully investigated. Many stakeholders are directly and indirectly involved in poaching activities (e.g., some local restaurants illegally providing meat to tourists). Therefore, stopping or severely decapitating the poaching industry requires a unified understanding of all stakeholders. The best way to uncover these geographical and social relationships is to track the movements of poachers. However, location tracking is challenging in most rural areas where wildlife sanctuaries are typically located. Internet-connected communication (e.g. 3G) technologies typically used in urban cities are not feasible in these rural areas. Therefore, we decided to develop an SMS (short message service) base low-cost tracking system (SMS-TRACCAR) to track poachers. The proposed system was developed to be deployed in Kinabatangan Wildlife Sanctuary, Sabah, Malaysia and nearby villages and cities where poachers typically move around. Our evaluations demonstrated that SMS-based tracking could provide sufficient quality (granular) data (with minimum energy consumption) that enable us to monitor poacher vehicle movements within rural areas where no other modern communication technologies are feasible to use. However, it is important to note that our system can be used in any domain that requires SMS-based geo-location tracking. SMS-TRACCAR can be configured to track individuals as well as groups. Therefore, SMS-TRACCAR contributes not only to the wildlife domain but in the wider context as well.

show abstract

“…1 This task aims to retrieve all the images of query vehicles in a large gallery set, where images are usually captured from nonoverlapping cameras under various viewpoints. Some previous works [2][3][4] demonstrated that the local-specific features, such as license plate and custom logos, can provide recognizable information for identifying two vehicles. However, due to the influence of illumination, occlusion, or cross-view in the real environment, subtle local-specific features might be unavailable.…”

Section: Introductionmentioning

confidence: 99%