arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0806.1446cs.CV
收藏

基于小波的快速视觉分类

Fast Wavelet-Based Visual Classification

Ecole Polytechnique(巴黎综合理工学院);Massachusetts Institute of Technology(麻省理工学院)

Guoshen Yu, Jean-Jacques Slotine

摘要:

我们研究了一种生物学动机的方法来快速视觉分类,直接受到Serre等人最近工作的启发。具体来说,权衡生物学准确性和计算效率,我们探索使用小波和grouplet样变换来并行调节视觉皮层V1和V2细胞,交替使用最大操作来实现尺度和平移不变性。在学习过程中应用特征选择过程以加速识别。我们引入了一个简单的注意力反馈机制,显着提高识别和鲁棒性在多对象场景。在实验中,该算法在目标识别、纹理和卫星图像分类、语言识别和声音分类上达到或超过了最先进的成功率。

英文摘要:

We investigate a biologically motivated approach to fast visual classification, directly inspired by the recent work of Serre et al. Specifically, trading-off biological accuracy for computational efficiency, we explore using wavelet and grouplet-like transforms to parallel the tuning of visual cortex V1 and V2 cells, alternated with max operations to achieve scale and translation invariance. A feature selection procedure is applied during learning to accelerate recognition. We introduce a simple attention-like feedback mechanism, significantly improving recognition and robustness in multiple-object scenes. In experiments, the proposed algorithm achieves or exceeds state-of-the-art success rate on object recognition, texture and satellite image classification, language identification and sound classification.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑