Deep Feature Augmentation for Occluded Image Classification

no code implementations • 2 Nov 2020 • Feng Cen, Xiaoyu Zhao, Wuzhuang Li, Guanghui Wang

To alleviate the dependency on large-scale occluded image datasets, we propose a novel approach to improve the classification accuracy of occluded images by fine-tuning the pre-trained models with a set of augmented deep feature vectors (DFVs).

Classification General Classification +1