A Deep Learning-Based Vehicle Re-identification Algorithm with Attention and Joint Learning

The objective of identifying the weight of vehicles typically involves pinpointing a specific vehicle among a vast array of vehicular data, a process that diverges from the typical categorization process. To achieve this goal of vehicle weight identification, this study develops an algorithm for recognizing vehicle weight that incorporates an attention mechanism, drawing upon principles of deep learning. The paper presents a method for vehicle re-identification by replicating the algorithm's model, utilizing Resnet50 as the foundational network architecture for training purposes. On top of the baseline model, an attention mechanism module is integrated, and the concurrent training of multiple loss functions is employed to enhance the model's performance. The effectiveness of the model is subsequently confirmed and appraised using a publicly available dataset. The experimental outcomes indicate that the model's performance can be enhanced by incorporating an attention mechanism, thereby successfully completing the task of vehicle weight identification. Post-simulation, the algorithm's precision is assessed under real-world road conditions.