Learning Consistency from High-quality Pseudo-labels for Weakly Supervised Object Localization
This work addresses localization accuracy in computer vision for tasks with limited supervision, representing an incremental improvement over existing pseudo-supervised methods.
The paper tackles the problem of weakly supervised object localization by proposing a two-stage approach that learns consistency from high-quality pseudo-labels, achieving excellent performance on benchmark datasets like CUB-200-2011, ImageNet-1k, and Tiny-ImageNet.
Pseudo-supervised learning methods have been shown to be effective for weakly supervised object localization tasks. However, the effectiveness depends on the powerful regularization ability of deep neural networks. Based on the assumption that the localization network should have similar location predictions on different versions of the same image, we propose a two-stage approach to learn more consistent localization. In the first stage, we propose a mask-based pseudo label generator algorithm, and use the pseudo-supervised learning method to initialize an object localization network. In the second stage, we propose a simple and effective method for evaluating the confidence of pseudo-labels based on classification discrimination, and by learning consistency from high-quality pseudo-labels, we further refine the localization network to get better localization performance. Experimental results show that our proposed approach achieves excellent performance in three benchmark datasets including CUB-200-2011, ImageNet-1k and Tiny-ImageNet, which demonstrates its effectiveness.