Lighting and Rotation Invariant Real-time Vehicle Wheel Detector based on YOLOv5

TL;DR

This paper presents a real-time vehicle wheel detector based on YOLOv5 that is invariant to lighting and rotation changes, addressing challenges in diverse imaging conditions without relying on extensive labeled datasets.

Contribution

The paper introduces a novel approach for creating lighting and rotation invariant vehicle wheel detection using YOLOv5, suitable for environments with limited labeled data.

Findings

Achieved robust detection under varying lighting conditions

Maintained high accuracy across different rotation angles

Demonstrated real-time performance suitable for practical applications

Abstract

Creating an object detector, in computer vision, has some common challenges when initially developed based on Convolutional Neural Network (CNN) architecture. These challenges are more apparent when creating model that needs to adapt to images captured by various camera orientations, lighting conditions, and environmental changes. The availability of the initial training samples to cover all these conditions can be an enormous challenge with a time and cost burden. While the problem can exist when creating any type of object detection, some types are less common and have no pre-labeled image datasets that exists publicly. Sometime public datasets are not reliable nor comprehensive for a rare object type. Vehicle wheel is one of those example that been chosen to demonstrate the approach of creating a lighting and rotation invariant real-time detector based on YOLOv5 architecture. The objective is to provide a simple approach that could be used as a reference for developing other types of real-time object detectors.