pptvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv125.pptx

gfdfhhjh

1. Department of Master of Computer Applications, Nitte Meenakshi Institute of Technology, Bengaluru, India By, Shama Bhat Sathish Ragini Shashank H R Soundarya R Enhancing Trolley Shopping with IoT and Computer Vision: A Secure Solution

2. Contents Introduction Literature Review Proposed System Proposed Methodology Test Process and Results Conclusion

3. Introduction Fig.1: Block Diagram of Smart Shopping Cart This research presents an IoT-enabled trolley that scans the products automatically using computer vision and integrates secured transaction mechanisms. The proposed system influences affordable and accessible IoT devices to create a smart shopping cart that can automatically identify and scans the products using computer vision techniques.

4. Literature Review IoT-Enabled Smart Trolleys Pradeepkumar et al. [1] proposed a smart shopping trolley integrated with IoT and a mobile application. This system streamlines shopping by automatically identifying products, updating the list in real-time, and enabling seamless checkout. Subudhi and Ponnalagu [2] introduce an intelligent shopping cart system featuring automatic product detection and a secure payment system. This solution uses RFID to identify products and integrates a payment gateway for swift, secure transactions, reducing checkout wait times.

5. Advanced Billing Systems and Payment Integration Shankar et al. [4] developed a smart trolley equipped with an advanced billing system using IoT. This system automates billing by scanning items in the trolley, using IoT for real-time data updates, enhancing accuracy and speed. Sarwar et al. [12] focus on smart shopping carts that utilize mobile computing and deep learning cloud services. These carts automate checkout with advanced data processing and cloud services, ensuring scalability and efficient data management. Zhang et al. [10] developed an IoT-based intelligent shopping cart to enhance efficiency and convenience.

6. Enhanced User Experience and Accessibility Rajkanna et al. [13] study the efficiency of smart trolleys in supermarkets, highlighting time savings, ease of use, and customer satisfaction. Siva Rao et al. [8] developed "Shop GO," an IoT-based solution simplifies shopping. Kanthimathi et al. [16] designed an AI-based trolley to assist visually impaired shoppers, enhancing accessibility. Yao et al. [15] introduced a radar self-following shopping cart utilizing multi-sensor fusion to autonomously navigate and assist shoppers. Table 1, concludes the discussion on the methods, results and advantages and disadvantages of the existing research.

7. The main contributions of this paper are: Using IoT to integrate security and real-time product scanning For accurate product recognition in trolleys, Faster Region-based Convolutional Neural Network (Faster R-CNN with LSTM) is used. The Long Short-Term Memory (LSTM) network, which manages temporal dependencies and sequence prediction.

8. The components of the proposed system is described as follows: Smart Trolley Design Computer Vision Module IoT Communication Module Security Module Central Store System

9. Proposed System Significant advancements in IoT-based shopping systems and smart carts aim to enhance the shopping experience through automation, secure payment methods, effective product identification, and improved customer accessibility. These systems integrate blockchain, RFID, computer vision, and IoT. Future research should focus on system integration, scalability, and user acceptance to improve effectiveness. The proposed IoT-enabled smart shopping cart leverages technologies like computer vision and secure payment systems to optimize shopping.

10. Fig.2: Block Diagram of proposed shopping Cart using IoT and Computer Vision

11. Proposed Methodology Experimental Setup The entire work for the smart shopping cart is implemented in python tool using computer vision algorithms for object detection using RFID/Barcode. IoT platform for connecting the trolley to the cloud based database, enabling data exchange and remote monitoring. The system is tested in a simulated environment to assess accuracy, response time, user satisfaction, and security. Dataset Description The dataset comprises high-resolution barcode images from a local supermarket, covering various retail products as presented in table 2. Each item is photographed from multiple angles, ensuring real-world robustness. Detailed metadata includes product details like names, categories, and prices. Barcode images and RFID tag readings aid accurate identification and inventory management. Data is split into training, validation, and test sets, with augmentation for generalization. Security measures anonymize personal data, ensuring ethical

12. smart trolleys using barcode/RFID is a hybrid model that combines CNNs Faster R-CNN with the sequence model LSTM. Proposed Faster R-CNN with LSTM for item detection in shopping cart When a product is inserted into the trolley, the procedure starts; the system takes a picture and looks for an RFID tag or barcode. The barcode/RFID tag in the picture is swiftly located and identified by the CNN component. The product details are retrieved from a central database by decoding the barcode/RFID information once it has been spotted. The sequence model component ensures that each product is appropriately logged by the system by linking scanned barcodes or RFID tags with the appropriate products.

13. This hybrid approach leverages the precise detection capabilities of CNNs and the sequential data handling strengths of LSTMs, enabling real-time and accurate item identification, reducing errors, and enhancing the shopping experience by automating inventory updates and billing processes. Fig.3: Faster-CNN with LSTM for Item detection

14. Test Process and Results Fig.3: Faster-CNN with LSTM model loss comparisons

15. Conclusion Implementing the Faster R-CNN-LSTM model in the IoT -enabled smart system for product scanning and secured trolley shopping proved highly effective. While results are promising, future work could reduce error rates with better occlusion handling and a more diverse dataset. Integrating advanced payment methods, such as biometric authentication could further enhance security and user convenience.

16. THANK YOU