Najib, Suhaila Mohd and Mirin, Siti Nur Suhaila and Harman, Ahmad Irfan and Mohd Rahimi, Muhammad Qarl Farisz and Rahim, Muhammad Daniel and Azhari, Nurnajihah Hazirah and Khang, Adam Wong Yoon (2023) Road hazard detection for the motorcycle based on efficientnet-lite0. In: 13th IEEE Symposium on Computer Applications and Industrial Electronics, ISCAIE 2023, 20 May 2023through 21 May 2023, Penang.
![[img]](http://eprints.utem.edu.my/style/images/fileicons/text.png) |
Text
Road hazard detection for the motorcycle based on efficientnet-lite0.pdf Restricted to Repository staff only Download (1MB) |
Abstract
With the limitation of shorter stopping distance, the motorcycle needs more space to break. The existence of static road hazards has a higher potential for road crashes which leads to a higher risk of serious injury to motorcyclists. Road hazards can be identified and located using the Motorcycle Object Detection system or simply known as MOD. The MOD system has the ability to detect and classify multiple objects in real-time. It employs the TensorFlow Lite framework on edge devices i.e., Raspberry Pi 4. TensorFlow Lite is the best preference for the Raspberry Pi 4 for deploying a pre-trained neural network; EfficientNet-Lite0 model. The MOD system utilizes an 8MP camera to capture the presence of the trained objects such as potholes, cones, barriers, and branches from the opposite direction which enter the motorcyclist's Region of Interest (RoI). RoI is designed based on the motorcyclist's point of view (30 meters ahead) which can reduce false object detection. Likewise, RoI can prevent excessive alert noise to the motorcyclist. MOD system serves to alert the motorcyclist to the presence of the trained objects and activates the audible and visual warning system through the built-in speaker in the helmet and the warning light installed on the handlebar of the motorcycle.
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Uncontrolled Keywords: | Efficientnet-lite0, Motorcycle object detection, Raspberry Pi 4, Tensorflow lite |
| Divisions: | Faculty Of Electronics And Computer Technology And Engineering |
| Depositing User: | Maizatul Najwa Ahmad |
| Date Deposited: | 20 Sep 2024 16:39 |
| Last Modified: | 20 Sep 2024 16:39 |
| URI: | http://eprints.utem.edu.my/id/eprint/27925 |
| Statistic Details: | View Download Statistic |
Actions (login required)
 |
View Item |