Deep Roto-Translation Scattering for Object Classification
This work addresses image classification by refining representations with geometric priors, showing incremental improvement over predefined features.
The paper tackled object classification by introducing a deep scattering convolution network with predefined wavelet filters over spatial and angular variables, achieving accuracy comparable to unsupervised deep learning and dictionary-based methods on datasets like Caltech and CIFAR.
Dictionary learning algorithms or supervised deep convolution networks have considerably improved the efficiency of predefined feature representations such as SIFT. We introduce a deep scattering convolution network, with predefined wavelet filters over spatial and angular variables. This representation brings an important improvement to results previously obtained with predefined features over object image databases such as Caltech and CIFAR. The resulting accuracy is comparable to results obtained with unsupervised deep learning and dictionary based representations. This shows that refining image representations by using geometric priors is a promising direction to improve image classification and its understanding.