Concurrently Extrapolating and Interpolating Networks for Continuous Model Generation
This work addresses efficiency issues in training image smoothing models for computer vision applications, though it appears incremental in nature.
The paper tackles the problem of costly repetitive training for deep image smoothing operators with different parameters by proposing a continuous model generation strategy that requires only one set of label images. The method achieves better performance than several state-of-the-art methods for image smoothing, as shown through objective and visual experimental results.
Most deep image smoothing operators are always trained repetitively when different explicit structure-texture pairs are employed as label images for each algorithm configured with different parameters. This kind of training strategy often takes a long time and spends equipment resources in a costly manner. To address this challenging issue, we generalize continuous network interpolation as a more powerful model generation tool, and then propose a simple yet effective model generation strategy to form a sequence of models that only requires a set of specific-effect label images. To precisely learn image smoothing operators, we present a double-state aggregation (DSA) module, which can be easily inserted into most of current network architecture. Based on this module, we design a double-state aggregation neural network structure with a local feature aggregation block and a nonlocal feature aggregation block to obtain operators with large expression capacity. Through the evaluation of many objective and visual experimental results, we show that the proposed method is capable of producing a series of continuous models and achieves better performance than that of several state-of-the-art methods for image smoothing.