Open Access

ARTICLE

Optimizing CNN Class Granularity for Power-Efficient Edge AI in Sudden Unintended Acceleration Verification

HeeSeok Choi¹, Joon-Min Gil^2,*

1 Elkanah Research and Development Center, Elkanah Corp, Seoul, Republic of Korea
2 Department of Computer Engineering, Jeju National University, Jeju, Republic of Korea

* Corresponding Author: Joon-Min Gil. Email:

(This article belongs to the Special Issue: Integrating Computing Technology of Cloud-Fog-Edge Environments and its Application)

Computers, Materials & Continua 2026, 87(2), 73 https://doi.org/10.32604/cmc.2026.074511

Received 13 October 2025; Accepted 15 January 2026; Issue published 12 March 2026

Abstract

Given the growing number of vehicle accidents caused by unintended acceleration and braking failure, verifying Sudden Unintended Acceleration (SUA) incidents has become a persistent challenge. A central issue of debate is whether such events stem from mechanical malfunctions or driver pedal misapplications. However, existing verification procedures implemented by vehicle manufacturers often involve closed tests after vehicle recalls; thus raising ongoing concerns about reliability and transparency. Consequently, there is a growing need for a user-driven framework that enables independent data acquisition and verification. Although previous studies have addressed SUA detection using deep learning, few have explored how class granularity optimization affects power efficiency and inference performance in real-time Edge AI systems. To address this problem, this work presents a cloud-assisted artificial intelligence (AI) solution for the reliable verification of SUA occurrences. The proposed system integrates multimodal sensor streams including camera-based foot images, On-Board Diagnostics II (OBD-II) signals, and six-axis measurements to determine whether the brake pedal was actually engaged at the moment of a suspected SUA. Beyond image acquisition, convolutional neural network (CNN) models perform real-time inference to classify the driver’s pedal operation states with the resulting outputs transmitted and archived in the cloud. A dedicated dataset of brake and accelerator pedal images was collected from 15 vehicles produced by 6 domestic and international manufacturers. Using this dataset, transfer learning techniques were applied to compare and analyze model performance and generalization as the CNN class granularity varied from coarse to fine levels. Furthermore, classification performance was evaluated in terms of latency and power efficiency under different class configurations. The experimental results demonstrated that the proposed solution identified the driver’s pedal behavior accurately and promptly, with the two-class model achieving the highest F1-score and accuracy among all granularity settings.

---

Keywords

---