Image Resizing by Reconstruction from Deep Features
This addresses image resizing challenges for computer vision applications by leveraging deep learning, though it is incremental as it builds on existing pre-trained networks.
The paper tackles image resizing by performing it in feature space using deep neural network features to preserve semantic content, resulting in reduced artifacts compared to traditional pixel-space methods.
Traditional image resizing methods usually work in pixel space and use various saliency measures. The challenge is to adjust the image shape while trying to preserve important content. In this paper we perform image resizing in feature space where the deep layers of a neural network contain rich important semantic information. We directly adjust the image feature maps, extracted from a pre-trained classification network, and reconstruct the resized image using a neural-network based optimization. This novel approach leverages the hierarchical encoding of the network, and in particular, the high-level discriminative power of its deeper layers, that recognizes semantic objects and regions and allows maintaining their aspect ratio. Our use of reconstruction from deep features diminishes the artifacts introduced by image-space resizing operators. We evaluate our method on benchmarks, compare to alternative approaches, and demonstrate its strength on challenging images.