Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jan 9, 2026

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

996

Deploying TinyML for energy-efficient object detection and communication in low-power edge AI systems.

Ch Madhu Bhushan1, Priya Koppuravuri1, Nomitha Prasanthi1

  • 1Department of CSE, SRM University AP, Amaravati, Andhra Pradesh, India.

Scientific Reports
|December 5, 2025
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Post-quantum distributed ledger technology: a systematic survey.

Scientific reports·2023
Same author

SONG: A Multi-Objective Evolutionary Algorithm for Delay and Energy Aware Facility Location in Vehicular Fog Networks.

Sensors (Basel, Switzerland)·2023
See all related articles

This study optimizes Edge Artificial Intelligence (Edge AI) for microcontrollers by using model compression techniques like quantization. The resulting system offers efficient, real-time object detection for IoT devices with minimal resource usage.

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Embedded Systems

Background:

  • Edge Artificial Intelligence (Edge AI) enables on-device processing for resource-constrained microcontrollers (MCUs).
  • Deploying deep learning models on MCUs faces challenges due to limited memory, computation, and energy.
  • Real-time, energy-efficient solutions are crucial for industrial automation, IoT, and smart devices.

Purpose of the Study:

  • To develop a real-time object detection system optimized for energy efficiency and scalability on MCUs.
  • To integrate model compression techniques, specifically quantization, with a lightweight neural network (MobileNetV2).
  • To analyze system-level trade-offs between latency, memory, energy consumption, and model size.

Main Methods:

  • Leveraged MobileNetV2, a lightweight neural network, and applied 8-bit post-training quantization.

Related Experiment Videos

Last Updated: Jan 9, 2026

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

996
  • Integrated a camera and Wi-Fi module with dual-mode TCP/UDP communication for data transmission.
  • Conducted system-level analysis using the Visual Wake Words (VWW) dataset on an MCU platform.

    • Main Results:

    • Achieved 3x storage reduction with 8-bit quantization, fitting within a 1MB flash/256KB SRAM budget (286-536 KB footprints).
    • On-device inference latency ranged from 3.47 to 14.98 ms per frame, with energy consumption of 10.6-22.1 J per inference.
    • Quantized MobileNet variants maintained accuracy, and UDP offered lower latency than TCP for wireless transmission.

    Conclusions:

    • The proposed system demonstrates a practical and scalable solution for real-time Edge AI on constrained hardware.
    • Integration of TinyML models with MCUs provides a foundation for autonomous, energy-efficient Edge AI.
    • Application-dependent protocol choices (TCP/UDP) are critical for balancing reliability and latency in embedded systems.