Xiaoyu Dong

The University of Tokyo | Todai

Neural network quantization aims at reducing bit-widths of weights and activations for memory and computational efficiency. Since a linear quantizer (i.e., round(·) function) cannot well fit the bell-shaped distributions of weights and activations, many existing methods use pre-defined functions (e.g., exponential function) with learnable parameter...

Self-supervised cross-modal super-resolution (SR) can overcome the difficulty of acquiring paired training data, but is challenging because only low-resolution (LR) source and high-resolution (HR) guide images from different modalities are available. Existing methods utilize pseudo or weak supervision in LR space and thus deliver results that are b...

Neural networks contain considerable redundant computation, which drags down the inference efficiency and hinders the deployment on resource-limited devices. In this paper, we study the sparsity in convolutional neural networks and propose a generic sparse mask mechanism to improve the inference efficiency of networks. Specifically, sparse masks ar...

Arbitrary-oriented object detection (AOOD) is widely used in aerial images because of its efficient object representation. However, current detectors employ the over-standardized feature extraction structure, resulting in detectors has no ability to adaptively readjust feature representations of detection units. Meanwhile, we observe that many dete...

Self-supervised cross-modal super-resolution (SR) can overcome the difficulty of acquiring paired training data, but is challenging because only low-resolution (LR) source and high-resolution (HR) guide images from different modalities are available. Existing methods utilize pseudo or weak supervision in LR space and thus deliver results that are b...

Most existing CNN-based super-resolution (SR) methods are developed based on an assumption that the degradation is fixed and known (e.g., bicubic downsampling). However, these methods suffer a severe performance drop when the real degradation is different from their assumption. To handle various unknown degradations in real-world applications, prev...

Remotely sensed images, especially in urban areas, have highly complex spatial distribution, since the ground objects have diverse ranges of sizes and shapes. This largely increases the difficulty of super-resolution (SR) tasks. Current deep convolutional neural network (CNN)-based SR methods often show limited performance when coping with complica...

Current CNN-based super-resolution (SR) methods process all locations equally with computational resources being uniformly assigned in space. However, since missing details in low-resolution images mainly exists in regions of edges and textures, less computational resources are required for those flat regions. Therefore, existing CNN-based methods...

Super-resolution (SR) techniques play a crucial role in increasing the spatial resolution of remote sensing data and overcoming the physical limitations of the spaceborne imaging systems. Though the convolutional neural network (CNN)-based methods have obtained good performance, they show limited capacity when coping with large-scale super-resolvin...

Image super-resolution (SR) reconstruction plays a key role in coping with the increasing demand on remote sensing imaging applications with high spatial resolution requirements. Though many SR methods have been proposed over the last few years, further research is needed to improve SR processes with regard to the complex spatial distribution of th...