International Journal of Advances in Applied Sciences (IJAAS) Vol.
No.
March 2026, pp.
ISSN: 2252-8814.
DOI: 10.
11591/ijaas.
Artificial intelligence-powered image recognition retail checkout systems Malyssa Alias1.
Dhaifina Saidi1.
Lim Jia Huey1.
Lee Qing Fang1.
Durghaashini S.
Ragunathan1.
JosephNg Poh Soon1.
Phan Koo Yuen2.
Lim Jit Theam2.
Wong See Wan3 Institute of Computer Science and Digital Innovation.
UCSI University.
Kuala Lumpur.
Malaysia Faculty of Information and Communication Technology.
Universiti Tunku Abdul Rahman.
Kampar.
Malaysia Vise Trading.
Penang.
Malaysia
Article Info
ABSTRACT
Article history:
The integration of artificial intelligence (AI) with big data analytics leads to substantial transformations in the retail sector.
This research explores the impact of AI-powered image recognition checkout systems on the retail industry, focusing on operational efficiency, customer experience, and resource waste.
Employing a mixed-methods approach, this study combines usability testing and data analytics to assess the viability of this technology in attaining automation and accuracy in retail operations.
The study focuses on the creation of robust, resource-efficient systems that foster long-term industrial growth.
The findings show that AI-powered solutions not only speed the checkout process but also contribute to sustainable infrastructure by reducing resource consumption and increasing energy efficiency.
This report offers significant information, like the impact of AI-powered image recognition checkout systems on operational efficiency, customer experience, and the role of AI in promoting sustainable infrastructure for retailers and governments looking to advance the digitalization of the retail Received Apr 8, 2025 Revised Dec 2, 2025 Accepted Jan 1, 2026 Keywords:
Artificial intelligence Big data Operational efficiency Retail innovation Sustainable infrastructure This is an open access article under the CC BY-SA license.
Corresponding Author:
JosephNg Poh Soon Institute of Computer Science and Digital Innovation.
UCSI University Jalan UCSI.
UCSI Heights (Taman Connaugh.
Cheras 56000.
Kuala Lumpur.
Malaysia Email: joseph.
ng@ucsiuniversity.
INTRODUCTION
The retail industry is undergoing a significant transformation, driven by technological advances and evolving consumer preferences.
Automation and artificial intelligence (AI) have become fundamental to these transformations, allowing businesses to streamline operations and improve customer experience .
To stay competitive, grocery stores are integrating automated technology into their current setups .
Automation significantly impacts financial performance, operational efficiency, and brand image.
AI-powered tools optimize inventory management, reduce labor costs, and improve decision-making processes .
Furthermore, cashier-free solutions minimize wait times, creating a smooth shopping experience that strengthens brand appeal .
Content marketing has also emerged as a pivotal strategy in the digital era, enabling organizations to share purposeful information that fosters long-term relationships with consumers while creating value through high-quality, engaging content .
The benefits include higher customer satisfaction through shortened checkout times, improved data analytics for inventory optimization, and increased market share .
However, issues such as high implementation costs, cybersecurity risks, and potential backlash over job displacement require strategic planning.
Budget restrictions and a lack of technical experience make it difficult for smaller merchants to embrace such technology .
Similarly, a thorough evaluation of Journal homepage: http://ijaas.
ISSN: 2252-8814
machine learning applications in retail is provided, focusing on prospects for predictive analytics and customer personalization .
These findings are consistent with the study of cashier-free checkout, emphasizing their potential to redefine in-store shopping experiences .
In this digital era, it is commonly known that technology is constantly progressing.
The core theme of this research revolves around the need to improve grocery store operations includes long checkout lines caused by manual checkouts.
One of the benefits includes improving efficiency.
Automating the checkout process ensures the system is prepared to track the customerAos session, eliminating the need for manual Moreover, adopting advanced AI systems positions the store as innovative and attracts tech-savvy customers.
The systemAos long-term advantages will make the initial cost expenditure worthwhile.
Current checkout systems still rely on traditional checkout systems with weaknesses, such as the inadequacy of inventory management.
They do not automatically update the product inventory in real-time .
This leads to inventory management issues and increases operational costs.
Furthermore, the current checkout systems solely depend on barcode scanners, which leads to hardware constraints, such as barcode issues.
Back-end databases are required for radio frequency identification (RFID) systems to maintain customer data, transaction records, and inventory information.
To enhance the customer shopping experience, we propose an AI-powered image recognition checkout system.
This technology replaces the traditional checkout system for customers to pick up items, pay online, and leave the store .
Self-checkout systems have become an essential component in retail environments, offering customers a more convenient way to complete their purchases .
These seamless systems help reduce the burden on staff and minimize wait times through advanced computer vision technology with image processing.
Strategically installing cameras to identify each item based on size, color, and packaging.
Image processing techniques will preprocess the captured images to ensure accuracy.
A deep learning model trained on large datasets improves recognition and streamlines checkout.
This benefits customers and retailers by eliminating long queues, boosting customer satisfaction, and increasing store AI is widely used in many fields besides retail such as in citizen science to improve species identification in biodiversity monitoring.
The researchers used a machine learning model integrated with a web interface that combined AI-generated predictions and visual feature keys .
As customers add things to their carts, the RFID reader scans the tags on the products and updates the total cost on the display .
Figure 1 shows the research framework for this analysis.
Figure 1.
Research framework The research hypothesis (RH) that is fundamental in this analysis is as follows:
RH1: AI-powered image recognition technology significantly contributes to the efficiency of checkout systems in retail industries.
This technology employs computer vision and deep learning to accurately identify products and calculate bills in real-time without barcode scanners and manual entry, before checkout for seamless transactions.
RH2: AI-powered image recognition checkout systems will enhance customer experience in the retail Based on research and capabilities of AI-powered image recognition systems in retail, we found that they will notably enhance the experience, efficiency, security, and trust by streamlining the checkout process.
RH3: AI-powered image recognition checkout systems help enhance decision-making and make operations more efficient in retail management.
This hypothesis assumes that AI checkout systems in retail enhance decision-making and operational performance by automating checkout processes and providing real-time data analytics.
By simplifying operations, these systems make it easier for better inventory management, staffing choices, and customer service, culminating in greater overall retail performance.
Int J Adv Appl Sci.
Vol.
No.
March 2026: 187-196
Int J Adv Appl Sci
ISSN: 2252-8814
RH4: Enhanced efficiency will lead to higher acceptance of AI-powered image recognition checkout systems within society in the retail industry.
This hypothesis proposes a positive relationship between efficiency and societal acceptance.
The survey will measure customer perceptions of efficiency and acceptance, along with statistical analysis to assess correlations.
RH5: AI-powered image recognition checkout systems help reduce resource waste and enhance sustainability in retail industries.
This research features a detailed exploration of key points relating to the use of AI-powered image recognition in retail.
The objective is to showcase a structured and well-organized background for our research, as shown in Figure 2.
Table 1 presents the existing technologies employed in traditional checkout systems and highlights their associated challenges and future research directions.
The comparison shows that manual inventory updates and barcode-based scanning processes often lead to slower operations and data Additionally, high labor dependence reduces efficiency and negatively impacts customer These limitations underscore the need for an automated.
AI-driven checkout system capable of real-time tracking and seamless customer interaction.
Figure 2.
Flowchart of literature review concepts Table 1.
Current technology process Current technology process Inventory updates Labor dependence Customer experience Issues in traditional checkout systems Lack of real-time inventory updates causes errors, overstock, or shortages.
Retailers update inventory at the end of the day or week, affecting product availability .
High reliance on cashiers increases labor costs.
Barcode scanners require frequent maintenance and software updates, adding to operational expenses .
Long queues and checkout delays due to barcode scanning issues, system glitches, and staff workload.
These inefficiencies frustrate customers, reducing satisfaction and potential sales .
Future research directions Explore AI-powered management for real-time tracking and synchronization.
Conduct a cost-benefit analysis comparing AI checkout systems with traditional methods over 5-10 years.
Study consumer attitudes, digital literacy levels, and AI adoption barriers to develop smoother transition strategies.
The retail sector has rapidly experimented with cashier-less and vision-based checkout systems over the last five years.
Commercial systems such as AmazonAos AuJust walk outAy and AlibabaAos Freshippo exemplify two deployment archetypes.
The high-density sensor fusion .
ameras, weight sensors, and trackin.
versus a hybrid mobile or vision-integrated workflow.
Operational analyses emphasize throughput gains, shrinkage reduction, and staff retraining costs .
, .
On the algorithmic front, real-time object detection continues to evolve.
The you only look once (YOLO) family and scalable detectors like EfficientDet remain popular for their favorable speed-vs-accuracy balance.
Recent robotics work by Gholami et al.
demonstrates near real-time human detection in IoT contexts using YOLO frameworks, illustrating that such architectures can be adapted to dynamic, latency-sensitive environments.
Table 2 provides a benchmarking comparison of existing checkout technologies across operational, cost, and scalability dimensions.
Barcode scanning remains reliable but labor-intensive, whereas RFID improves automation but introduces cost and read-rate challenges.
Mobile self-checkout offers flexibility but relies heavily on customer honesty and digital literacy.
Vision-based checkout is supported by computer vision and deep learning, enabling a fully automated checkout experience.
Hybrid sensor fusion approaches currently offer the best balance between automation accuracy and real-time performance by combining cameras with weight sensors.
RFID, or optical character recognition (OCR)-based validation .
Ae.
This comparison highlights a research gap in optimizing vision-based systems for cost efficiency and customer acceptance in emerging markets such as Malaysia, which this study aims to address.
Artificial intelligence-powered image recognition retail checkout systems (Malyssa Alia.
A ISSN: 2252-8814 Table 2.
Benchmarking comparison Technology Mobile self-checkout can and g.
Vision-based checkout (AI camera.
Hybrid sensor fusion (AI scale/RFID/OCR) Advantages Ae Low implementation Ae Customer autonomy Ae Reduces cashier Ae Hands-free checkout Ae Fast throughput Ae Rich analytics .
ehavioral and shelf Ae Highest accuracy Ae Reduces occlusion Ae Improves recognition Limitations Ae Error-prone .
Ae High theft risk Ae Depends on customer digital literacy Cost and infrastructure Low .
obile app QR Suitable use cases Ae Retailers with budget limits Ae Hybrid Ae High initial cost Ae Complex deployment Ae Privacy concerns Ae Trained model Ae Requires multimodal Ae Higher maintenance High ameras edge/cloud AI) Ae Fully automated Ae Seamless retail Very high .
ensor combinations AI) Ae Amazon Go-style Ae High-demand
RESEARCH METHOD
This academic research adopted a mixed-method approach, combining empirical data collection with conceptual system development.
The first component involved developing questionnaires, observations, and interviews to evaluate AI-powered image recognition checkout systems in retail industries .
Ae.
other words, it utilized quantitative (Likert-scal.
and qualitative .
pen-ende.
survey questions to assess the impacts .
To ensure reliability and validity, this research used established tools, like Google Scholar, adapted to the topic.
Educational experts reviewed and finalized the questionnaires, which were distributed via Google Forms to collect survey data.
The target respondents were retail industry visitors, with convenience sampling gathering responses from at least 106 participants.
A Likert scale was used to maintain Figure 3 summarizes the questionnaire process.
Figure 3.
Flowchart of questionnaire observation MODEL ARCHITECTURE AND TRAINING The proposed product recognition system is designed using a deep learningAebased image classification framework intended to operate efficiently within retail checkout environments.
The conceptual architecture follows the principles of a convolutional neural network (CNN).
CNN is widely utilized for image recognition tasks due to its strong feature extraction and spatial pattern recognition capabilities .
Model architecture overview The planned model architecture consists of three key components, as follows:
Input layer: processes RGB product images resized to 640y640 pixels.
Feature extraction: convolutional layers with rectified linear unit (ReLU) and pooling capture key visual features while reducing computational load.
Classification and output: fully connected layers will map extracted features to their corresponding product categories, with the output layer generating a label and confidence score for each image.
Int J Adv Appl Sci.
Vol.
No.
March 2026: 187-196
Int J Adv Appl Sci
ISSN: 2252-8814
Training strategy The model will be trained on the dataset using the Adam optimizer .
earning rate 0.
for 50 epochs with early stopping.
Data augmentation .
, brightness changes, occlusion, and perspective shift.
will simulate real-world conditions.
Validation metrics will guide hyperparameter tuning.
Expected performance and future work The CNN is expected to deliver high accuracy and stable real-time performance at checkout Future enhancements may include transfer learning using architectures like MobileNet or EfficientNet to boost accuracy and reduce training time .
The conceptual CNN architecture processes images sequentially, starting with input, passing through convolutional and pooling layers, then flattening, and finally fully connected layers that output predicted product label and confidence score, as in Figure 4.
Figure 4.
Conceptual CNN architecture for product recognition EVALUATION PROTOCOL The analytical evaluation of this research aims to interpret the theoretical performance of the proposed AI-powered image recognition checkout framework.
Test scenarios To ensure a comprehensive assessment, the proposed system is evaluated across several simulated test scenarios derived from prior retail vision studies.
These include:
Single-product detection: evaluates recognition accuracy for one product per frame under different lighting and background conditions.
Multi-product basket scenario: tests the modelAos capability to detect multiple items placed closely together or partially overlapping .
cclusion handlin.
Dynamic customer interaction: simulates real-time movement as items are placed into or removed from the basket, to evaluate latency and frame-to-frame consistency.
Confusion matrix and performance metrics Performance evaluation would be guided by standard classification and detection metrics widely adopted in computer vision research.
A confusion matrix would be generated to illustrate the relationship between predicted and actual product classes, highlighting true positives, false positives, and false negatives.
From this matrix, several key metrics would be computed conceptually based on literature benchmarks .
Accuracy (Top-1/Top-.
: proportion of correctly recognized products out of total inputs.
Precision and recall: indicators of model reliability in identifying correct product categories while minimizing false detections.
F1-score: the metric used that combines both precision and recall.
Real-world validation .
heoretical analysi.
In terms of practical usability, the proposed systemAos workflow from image capture to automatic billing was assessed conceptually against practical deployment factors:
Environmental variability: lighting conditions, product overlap, and background textures consistent with those in the dataset description.
Operational latency: expected time delay between item detection and checkout confirmation, based on comparable edge-AI implementations .
User interaction flow: customer experience in a counter-free, self-service checkout, ensuring intuitive interface navigation and secure payment integration.
Scalability and system integration: potential challenges in connecting the recognition model with cloud inventory databases and pos systems.
This theoretical validation approach ensures that the proposed system design is realistic, implementable, and consistent with current Augrab-and-goAy retail technologies.
Future work would involve empirical validation through prototype testing.
It would also include confusion matrix generation and real-world benchmarking to substantiate the analytical findings presented here.
Artificial intelligence-powered image recognition retail checkout systems (Malyssa Alia.
A ISSN: 2252-8814
FINDINGS AND ANALYSIS
The survey achieved 106 respondents.
Quantitative answers are examined through descriptive statistics, and qualitative data is thematically analyzed to determine principal insights.
The survey, which was inspired by existing articles, takes about 10 minutes to complete and is formatted into demographic questions, scaled ratings of efficiency, customer experience, and sustainability, along with open-ended feedback .
, .
Figure 5 shows the response trends across theoretical, practical, and managerial knowledge areas.
The positive trend across all areas highlights the perceived effectiveness of AI-powered checkout systems in theoretical understanding, practical implementation, and managerial operations.
In the theoretical aspects, many respondents believe this system could impact the efficiency and speed of the checkout process by improving wait time and enhancing precision.
This indicates that the public believes this system will contribute positively to offering a seamless checkout system.
In the practical aspect, most respondents acknowledged that this implementation enhances the overall shopping experience as it enables faster Lastly, in the managerial aspect, many respondents believe that this system can improve the management of retail industries as it optimizes inventory tracking and increases efficiency in operation, improving overall store efficiency.
These inputs suggest a grasp of the potential for AI to improve operational performance and the shopping experience by optimizing resource use.
The high levels of agreement in both societal and sustainable aspects can be explained by the perception that AI image recognition systems enhance productivity and simplify work processes, and optimize operational efficiency, making them advantageous for practical and sustainable purposes.
In the societal aspects, many respondents mentioned factors such as data privacy, seamless integration, and cost as considerations that influenced them to accept this AI checkout system.
The societal approval indicates that participants believe these systems are useful in increasing productivity and making life easier, which may obscure ethical concerns for many of them.
In the sustainability aspect, respondents acknowledged this system's ability to reduce paper waste from receipts, minimize packaging waste, and improve energy These factors indicate an understanding of the potential of using AI to enhance the effectiveness of operations and reduce the negative impact on the environment through the optimal utilization of resources.
Figure 6 shows the response trends across societal and sustainable knowledge areas.
Figure 5.
Response trends across theoretical, practical, and managerial knowledge areas Figure 6.
Response trends across societal and sustainable knowledge areas Int J Adv Appl Sci.
Vol.
No.
March 2026: 187-196
Int J Adv Appl Sci
ISSN: 2252-8814
The contributions in Figure 7 manage to drive the widespread acceptance and integration of AI within contemporary retailing.
Beyond the systemAos current capabilities, future integration possibilities could further strengthen its impact on the retail ecosystem.
Connecting the AI-powered checkout system with the point-of-sale (POS) infrastructure enables seamless data synchronization and faster transaction processing.
Integrating with customer loyalty programs allows automatic reward updates during purchases, enhancing user engagement and retention.
In addition, linking the system to real-time analytics dashboards provides store managers with actionable insights into sales trends, customer preferences, and inventory movements.
These integrations would not only streamline retail operations but also empower data-driven decision-making that improves overall efficiency and customer satisfaction.
Figure 7.
Main contributions CONCLUSION This study demonstrates that AI-powered image recognition checkout systems significantly enhance retail efficiency by streamlining transactions, optimizing inventory management, and improving customer satisfaction through seamless integration.
The findings emphasize that successful implementation requires balancing technological innovation with strong privacy safeguards, including data minimization and regulatory compliance to maintain consumer trust.
Moreover, the research highlights the systemAos potential for continuous evolution through adaptive learning, multimodal recognition, and augmented reality integration, offering a strategic path toward a more intelligent, sustainable, and customer-centric retail ecosystem.
FUNDING INFORMATION
Authors state no funding involved.
Artificial intelligence-powered image recognition retail checkout systems (Malyssa Alia.
A ISSN: 2252-8814
AUTHOR CONTRIBUTIONS STATEMENT
This journal uses the Contributor Roles Taxonomy (CRediT) to recognize individual author contributions, reduce authorship disputes, and facilitate collaboration.
Name of Author Malyssa Alias Dhaifina Saidi Lim Jia Huey Lee Qing Fang Durghaashini S.
Ragunathan JosephNg Poh Soon Phan Koo Yuen Lim Jit Theam Wong See Wan C : Conceptualization M : Methodology So : Software Va : Validation Fo : Formal analysis ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue I : Investigation R : Resources D : Data Curation O : Writing - Original Draft E : Writing - Review & Editing ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue ue Vi : Visualization Su : Supervision P : Project administration Fu : Funding acquisition CONFLICT OF INTEREST STATEMENT Authors state no conflict of interest.
INFORMED CONSENT
We have obtained informed consent from all individuals included in this study.
DATA AVAILABILITY
Data availability is not applicable to this paper as no new data were created or analyzed in this study.
REFERENCES