Image-Based Vehicle Classification by Synergizing Features from Supervised and Self-Supervised Learning Paradigms

Abstract

This paper introduces a novel approach to leveraging features learned from both supervised and self-supervised paradigms, to improve image classification tasks, specifically for vehicle classification. Two state-of-the-art self-supervised learning methods, DINO and data2vec, were evaluated and compared for their representation learning of vehicle images. The former contrasts local and global views while the latter uses masked prediction on multiple layered representations. In the latter case, supervised learning is employed to finetune a pretrained YOLOR object detector for detecting vehicle wheels, from which definitive wheel positional features are retrieved. The representations learned from these self-supervised learning methods were combined with the wheel positional features for the vehicle classification task. Particularly, a random wheel masking strategy was utilized to finetune the previously learned representations in harmony with the wheel positional features during the training of the classifier. Our experiments show that the data2vec-distilled representations, which are consistent with our wheel masking strategy, outperformed the DINO counterpart, resulting in a celebrated Top-1 classification accuracy of 97.2% for classifying the 13 vehicle classes defined by the Federal Highway Administration.

Salient object detection: a mini review

“…Identifying salient areas in an image can facilitate subsequent advanced visual tasks, enhancing efficiency and resource management and improving performance (Gupta et al, 2020). Thus, SOD can help filter irrelevant backgrounds, and SOD plays a significant pre-processing role in computer vision applications, providing important basic processing for these applications, e.g., segmentation (Donoser et al, 2009;Qin et al, 2014;Noh et al, 2015;Fu et al, 2017;Shelhamer et al, 2017), classification (Borji and Itti, 2011;Joseph et al, 2019;Akila et al, 2021;Liu et al, 2021;Jia et al, 2022;Ma and Yang, 2023), tracking (Frintrop and Kessel, 2009;Su et al, 2014;Ma et al, 2017;Lee and Kim, 2018;Chen et al, 2019), etc.…”