Fine-Grained Vehicle Classification with Unsupervised Parts Co-Occurrence Learning
Abstract
Vehicle fine-grained classification is a challenging research problem with little attention in the field. In this paper, we propose a deep network architecture for vehicles fine-grained classification without the need of parts or 3D bounding boxes annotation. Co-occurrence layer (COOC) layer is exploited for unsupervised parts discovery. In addition, a two-step procedure with transfer learning and fine-tuning is utilized. This enables us to better fine-tune models with pre-trained weights on ImageNet in some layers while having random initialization in some others. Our model achieves 86.5% accuracy outperforming the state of the art methods in BoxCars116K by 4%. In addition, we achieve 95.5% and 93.19% on CompCars on both train-test splits, 70-30 and 50-50, outperforming the other methods by 4.5% and 8% respectively.
Cite
Text
Elkerdawy et al. "Fine-Grained Vehicle Classification with Unsupervised Parts Co-Occurrence Learning." European Conference on Computer Vision Workshops, 2018. doi:10.1007/978-3-030-11018-5_54Markdown
[Elkerdawy et al. "Fine-Grained Vehicle Classification with Unsupervised Parts Co-Occurrence Learning." European Conference on Computer Vision Workshops, 2018.](https://mlanthology.org/eccvw/2018/elkerdawy2018eccvw-finegrained/) doi:10.1007/978-3-030-11018-5_54BibTeX
@inproceedings{elkerdawy2018eccvw-finegrained,
title = {{Fine-Grained Vehicle Classification with Unsupervised Parts Co-Occurrence Learning}},
author = {Elkerdawy, Sara and Ray, Nilanjan and Zhang, Hong},
booktitle = {European Conference on Computer Vision Workshops},
year = {2018},
pages = {664-670},
doi = {10.1007/978-3-030-11018-5_54},
url = {https://mlanthology.org/eccvw/2018/elkerdawy2018eccvw-finegrained/}
}