2507.16114

#1 Stop-band Energy Constraint for Orthogonal Tunable Wavelet Units in Convolutional Neural Networks for Computer Vision problems [PDF¹] [Copy] [Kimi²] [REL]

Authors: An D. Le, Hung Nguyen, Sungbal Seo, You-Suk Bae, Truong Q. Nguyen

This work introduces a stop-band energy constraint for filters in orthogonal tunable wavelet units with a lattice structure, aimed at improving image classification and anomaly detection in CNNs, especially on texture-rich datasets. Integrated into ResNet-18, the method enhances convolution, pooling, and downsampling operations, yielding accuracy gains of 2.48% on CIFAR-10 and 13.56% on the Describable Textures dataset. Similar improvements are observed in ResNet-34. On the MVTec hazelnut anomaly detection task, the proposed method achieves competitive results in both segmentation and detection, outperforming existing approaches.

Subjects: Computer Vision and Pattern Recognition , Signal Processing

Publish: 2025-07-21 23:57:03 UTC

---