Automatic Identifier of Socket for Electrical Vehicles Using SWIN-Transformer and SimAM Attention Mechanism-Based EVS YOLO

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F23%3A10253574" target="_blank" >RIV/61989100:27240/23:10253574 - isvavai.cz</a>

Výsledek na webu

<a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10268929" target="_blank" >https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10268929</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/ACCESS.2023.3321290" target="_blank" >10.1109/ACCESS.2023.3321290</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Automatic Identifier of Socket for Electrical Vehicles Using SWIN-Transformer and SimAM Attention Mechanism-Based EVS YOLO

Popis výsledku v původním jazyce

Electric vehicle (EV) technology is emerging as one of the most promising solutions for green transportation. The same growth occurs in the charging infrastructure development and automating the EV charging process. Globally, EVs has different types of charging sockets and it&apos;s located at the various positions in the Vehicle. In simple, EV has a diversity in socket type and socket location. Hence, correctly identifying the socket type and location is mandatory to automate the charging process. The recent development in computer vision and robotic systems helps to automate EV charging without human intervention. Image processing and deep learning-based socket identification can help the EV charging infrastructure providers automate the process. Moreover, the deep learning techniques should be simple enough to implement in the real-time processing boards for experimental viability. Hence, this paper proposes a new You Only Look Once (YOLO) model called the Electric Vehicle Socket (EVS) YOLO that uses YOLOv5 as its base architecture with the addition of a vision-type transformer called the SWIN-Transformer and an attention mechanism called SimAM for better performance of the model in detecting the correct charging port. A dataset of 2700 images with six types of classes has been used to test the model, and the EVS -YOLO also evaluated with varying mechanisms of attention positioned at various places along the head. The paper contrasts the suggested model with alternative deep learning architectures and analyzes respective performances.

Název v anglickém jazyce

Automatic Identifier of Socket for Electrical Vehicles Using SWIN-Transformer and SimAM Attention Mechanism-Based EVS YOLO

Popis výsledku anglicky

Electric vehicle (EV) technology is emerging as one of the most promising solutions for green transportation. The same growth occurs in the charging infrastructure development and automating the EV charging process. Globally, EVs has different types of charging sockets and it&apos;s located at the various positions in the Vehicle. In simple, EV has a diversity in socket type and socket location. Hence, correctly identifying the socket type and location is mandatory to automate the charging process. The recent development in computer vision and robotic systems helps to automate EV charging without human intervention. Image processing and deep learning-based socket identification can help the EV charging infrastructure providers automate the process. Moreover, the deep learning techniques should be simple enough to implement in the real-time processing boards for experimental viability. Hence, this paper proposes a new You Only Look Once (YOLO) model called the Electric Vehicle Socket (EVS) YOLO that uses YOLOv5 as its base architecture with the addition of a vision-type transformer called the SWIN-Transformer and an attention mechanism called SimAM for better performance of the model in detecting the correct charging port. A dataset of 2700 images with six types of classes has been used to test the model, and the EVS -YOLO also evaluated with varying mechanisms of attention positioned at various places along the head. The paper contrasts the suggested model with alternative deep learning architectures and analyzes respective performances.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20200 - Electrical engineering, Electronic engineering, Information engineering

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

IEEE Access

ISSN

2169-3536

e-ISSN

—

Svazek periodika

11

Číslo periodika v rámci svazku

1

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

17

Strana od-do

111238-111254

Kód UT WoS článku

001086204500001

EID výsledku v databázi Scopus

2-s2.0-85174827300