Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM68211
Volume:
10 |
10 |
Oct 2026
Pages:
1-5
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Department of Computer Science and Engineering
GH Raisoni University Amravity
Department of Computer Science and Engineering
GH Raisoni University Amravity
Department of Computer Science and Engineering
GH Raisoni University Amravity
Department of Computer Science and Engineering
GH Raisoni University Amravity
Department of Computer Science and Engineering
GH Raisoni University Amravity
Retail shops, pharmacies, and small warehouses remain highly vulnerable to theft, unauthorized entry, and after-hours intrusion. Conventional Closed-Circuit Television (CCTV) systems are fundamentally passive — they record footage but cannot automatically distinguish an authorized employee from an intruder, analyze the nature of a threat, or notify the owner in real time. This paper presents an IoT-Based Smart Shop Monitoring System that fuses low-cost edge hardware, in-browser facial recognition, and multimodal generative AI into a single proactive security pipeline. The edge node combines an AI-Thinker ESP32-CAM, a Passive Infrared (PIR) motion sensor, and a 16×2 I²C LCD. On motion detection, the system captures an image, performs 128-dimensional facial verification locally (TinyFaceDetector + ResNet-34), and escalates suspicious frames to the Google Gemini Vision API for zero-shot threat classification (masks, weapons, concealment, after-hours presence). Confirmed threats trigger a 5-second WebM evidence recording and an instant Telegram alert containing the image, AI analysis, and evidence link. Experimental evaluation shows a PIR detection latency below 20 ms, 97.5% facial verification accuracy (AUC = 0.988), 95.8% Gemini threat-classification accuracy, and a mean end-to-end alert delivery time of 1.98 seconds. The results confirm that intelligent, automated, low-bandwidth shop security is achievable with affordable hardware.
KEYWORDS
This work designed, implemented, and validated a proactive, edge-cloud IoT shop security system that overcomes the passivity of conventional CCTV. Key contributions include a sub-20 ms ultra-low-power edge node, a 97.5%-accurate local biometric pipeline, 96% multimodal Gemini threat classification, automated 5-second evidence dispatch, and a zero-maintenance Auto-DDL SQLite backend — all delivering alerts in 1.98 seconds.
[1] D. M. Kharat, P. M. Jawandhiya, and S. M. Dandge, "Remotely Monitored Smart Security System," IJSRD, vol. 9, no. 4, pp. 453–455, 2021. [2] T. V. N. Jayudu, S. Nikhita, K. Syamala Bai, K. Sowdhanya, and R. C. Saranya, "IoT Based Anti-Theft Security System," JETIR, vol. 10, no. 4, Apr. 2023, paper ID JETIR2304529. [3] P. Syam Sai, MD. Ibrahim, and Sk. A. Siddiq, "Smart Security Surveillance System Using IoT," VFSTR, ECE Department, project report, 2024. [4] S. L. Nalbalwar, S. M. Fule, D. P. Jayaswal, and U. D. Meshram, "IoT-Based Smart Energy Meter with Real-Time Monitoring and Theft Detection using Arduino and PZEM004T," IJERT, Dec. 2025. [5] N. Imamah, S. WP, and A. Reynaldi, "Design and Implementation of an IoT-Based Store Security Monitoring System Using the ESP32-CAM Microcontroller," COMPUTING: Jurnal Informatika, vol. 9, no. 2, pp. 70–79, 2022. [6] D. Ramadhan, A. Hakim, and D. R. Irawati, "Sistem Pemantauan dan Keamanan pada Toko Berbasis Mikrokontroler ESP32 DevKit V1, ESP32-CAM AI Thinker, Sensor AM312 dan Buzzer Berbasis IoT," Journal of Information System, Informatics and Computing, vol. 9, no. 1.
Publication Date
10/22/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
6
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...